Active Galactic Nuclei: Masses and Dynamics
Dissertation
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University
By
Catherine J. Grier
Graduate Program in Astronomy
The Ohio State University 2013
Dissertation Committee: Professor Bradley M. Peterson, Advisor Professor Richard W. Pogge Professor Paul Martini Copyright by
Catherine J. Grier
2013 ABSTRACT
In this dissertation, I present the results of work to improve measurements of black hole masses (MBH) and investigate the structure of the broad line region
(BLR) in active galactic nuclei (AGNs), which are key tracers of galaxy formation and evolution. I address these issues by carrying out high-cadence reverberation mapping experiments on several targets using the telescopes at MDM Observatory on Kitt Peak and by further testing the relationship between MBH and host-galaxy velocity dispersion (σ∗) as a calibrator for the MBH scale.
Reverberation mapping can be employed for two specific purposes: 1)
Measurement of MBH in AGNs and 2) Determination of BLR dynamics. We obtained improved reverberation measurements of the radius of the BLR and MBH for the object PG 2130+099 and performed a re-analysis of previous data. Reverberation data were also collected at several observatories over a 140-day span in 2010.
We obtained high sampling-rate light curves for five objects, from which we have measured the average radius of the Hβ-emitting region and calculated MBH. Our new measurements substantially improve previous measurements for these objects. We also measure the lag in the He ii λ4686 emission line relative to the optical continuum
ii in Mrk 335, which is the first robust lag measurement for a high-ionization line in a narrow-line Seyfert 1 galaxy.
We also recovered velocity-resolved reverberation lags for all five AGNs and velocity-delay maps for four. In all four objects, velocity-delay maps show evidence for gravitational infall and in some cases Keplerian motion. These maps constitute a large increase in the number of objects for which we have resolved velocity-delay maps and provide evidence supporting the reliability of reverberation-based MBH measurements.
Finally, I discuss the results of an investigation with the Gemini North 8-m telescope to obtain σ∗ measurements in high-mass quasars. I report successful measurements for four objects that improve the quality of two measurements near the high-mass end of the distribution and add two new objects to the sample. We update the entire AGN MBH–σ∗ sample and obtain a new estimate of the virial factor used to calibrate all MBH measurements in AGNs.
iii Dedicated to my family, Boyce, Cindy, and Chris Grier, and my boyfriend Thomas Beatty. I could not have done this without all of
your love and support.
iv ACKNOWLEDGMENTS
First, I’d like to thank my advisor, Brad Peterson, for all of his help and support. He has provided advice and friendship throughout my entire time here and is truly invested in the success and happiness of his graduate students, for which I am endlessly grateful. I’d also like to thank my committee members Richard Pogge and Paul Martini for their invaluable help with all aspects of running the enormous observing programs that my work entails.
I would like to thank my parents for their constant, unwavering support in this endeavor and in all of my endeavors in life. I’d also like to thank my brother Chris, who has been a role model of mine and a good friend for my entire life — he keeps me in line.
I would next like to thank my boyfriend, Thomas Beatty, for his love and support during the past few years. He has helped get me through some of the rougher periods, and like my family, he has been a source of unwavering support and strength during my time here.
I would also like to thank my roommate, Jill Gerke, for keeping me sane and for being the best roommate I’ve ever had for four years during this process. She
v (and the cats) gave me a place to relax and enjoy when I returned from the office, and for that I am eternally grateful. I also thank my various officemates through the years for making work at the office an entertaining and enjoyable experience.
The OSU department and the graduate student body as a whole deserves a thanks in this regard, as the people here are all wonderful people and the collaborative, friendly environment they provide is the reason why our program is so good.
I extend a special thanks to David Will, our computer tech, for saving me on numerous occasions and making my computing life much easier than I have any right to expect it to be. I thank all of my coauthors and collaborators on the various publications for their contributions to the various projects. I’d also like to thank the various OSU professors throughout the years who have gone out of their way to answer my questions and help out, particularly Todd Thompson, Chris Kochanek,
Smita Mathur, and Scott Gaudi.
vi VITA
2007 ...... B.S. Physics & Astronomy, The University of Illinois at Urbana-Champaign
2007 – 2009 ...... Graduate Teaching and Research Associate, The Ohio State University
2009 – 2012 ...... Head Teaching Associate, The Ohio State University
2010 ...... M.S. Astronomy, The Ohio State University
2013 ...... Presidential Fellow, The Ohio State University
PUBLICATIONS
Research Publications
1. Ganguly, R., Misawa, T., Lynch, R., Charlton, J. C., Eracleous, M., Hawthorn, M. J., and Grier, Catherine J., “Quasar Intrinsic Absorption in the HST Archive”, ASP Conference Series, 373, 297, (2007).
2. Wilhite, Brian C., Brunner, R. J., Grier, Catherine J., Schneider, D. P., & vanden Berk, D. E., “On the variability of quasars: a link between the Eddington ratio and optical variability?”, MNRAS, 383, 1232, (2008).
3. Grier, Catherine J., and 16 coauthors, “The Mass of the Black Hole in the Quasar PG2130+099”, ApJ, 688, 837, (2008).
4. Grier, Catherine J., and 10 coauthors, “Investigating the High-Luminosity End of the Active Galaxy MBH–σ∗ Relation”, Co-Evolution of Central Black Holes and Galaxies, IAU Symposium 267, (2009).
vii 5. Grier, Catherine J., Mathur, S., Ghosh, H., & Ferrarese, L., “Discovery of Nuclear X-ray Sources in SINGS Galaxies”, ApJ, 731, 60, (2011).
6. Grier, Catherine J., Peterson, B. M., Denney, K. D., Bentz, M.C, & Pogge, R. W., “New Results in Reverberation Mapping”, in Narrow-Line Seyfert 1 Galaxies and their Place in the Universe, Proceedings of Science, POS(NLS1), 52, (2011).
7. Assef, R. J., and 32 coauthors, including Grier, Catherine J., “Black Hole Mass Estimates Based on CIV are Consistent with Those Based on the Balmer Lines”, ApJ, 742, 93, (2011).
8. Grier, Catherine J., and 38 coauthors, “A Reverberation Lag for the High-ionization Component of the Broad-line Region in the Narrow-line Seyfert 1 Mrk 335”, ApJL, 744, 4, (2012).
9. Assef, R. J., Frank, S., Grier, Catherine J., Kochanek, C. S., Denney, K. D., & Peterson, B. M., “The Importance of Broad Emission-Line Widths in Single Epoch Black Hole Mass Estimates”, ApJ, 753L, 2, (2012).
10. Grier, Catherine J., and 42 coauthors, “Reverberation Mapping Results for Five Seyfert 1 Galaxies”, ApJ, 755, 60, (2012).
11. Dietrich, M., Peterson, B. M., Grier, Catherine J., and 7 coauthors, “Optical Monitoring of the Broad-Line Radio Galaxy 3C 390.3”, ApJ, 757, 53, (2012).
12. Grier, Catherine J., and 40 coauthors, “The Structure of the Broad Line Region in Active Galactic Nuclei: I. Reconstructed Velocity-Delay Maps”, ApJ, 764, 47, (2013).
13. Bentz, M.C., Denney, K. D., Grier, Catherine J., & 14 coauthors, “The Low-Luminosity End of the Radius-Luminosity Relationship for Active Galactic Nuclei”, ApJ, 767, 149, (2013).
14. Somers, G., Mathur, S., Martini, P., Watson, L. C., Grier, Catherine J., & Ferrarese, L., “Discovery of a Large Population of Ultraluminous X-ray Sources in the Bulge-less Galaxies NGC 337 and ESO 501-23”, ApJ, submitted, (2013).
15. Grier, Catherine J., and 9 coauthors, “Stellar Velocity Dispersion Mea- surements in High-Luminosity Quasar Hosts and Implications for the AGN Black Hole Mass Scale”, ApJ, 772, 2, (2013).
viii FIELDS OF STUDY
Major Field: Astronomy
ix Table of Contents
Abstract ...... ii
Dedication ...... iv
Acknowledgments ...... v
Vita ...... vii
List of Tables ...... xiv
List of Figures ...... xvii
Chapter 1 Introduction ...... 1
1.1 Supermassive Black Holes in Galaxies ...... 1
1.2 ActiveGalacticNuclei ...... 2
1.3 ReverberationMapping...... 5
1.4 Velocity-Resolved Reverberation Mapping ...... 7
1.5 The AGN MBH–σ∗ Relation ...... 8
1.6 ScopeofthisDissertation ...... 11
Chapter 2 The Mass of the Black Hole in the Quasar PG 2130+099 14
2.1 Background ...... 14
2.2 ObservationsandDataAnalysis ...... 15
x 2.2.1 Observations ...... 15
2.2.2 LightCurves ...... 16
2.3 TimeSeriesAnalysis ...... 18
2.3.1 Time delay measurements ...... 18
2.3.2 Line width measurement and mass calculations ...... 21
2.4 Discussion...... 22
Chapter 3 Reverberation Mapping Results for Five Seyfert 1 Galaxies 38
3.1 Background ...... 38
3.2 Observations...... 39
3.2.1 Spectroscopy ...... 39
3.2.2 Photometry ...... 41
3.3 LightCurves...... 42
3.3.1 SpectroscopicLightCurves...... 42
3.3.2 PhotometricLightCurves ...... 43
3.3.3 CombinedLightCurves ...... 44
3.4 TimeSeriesMeasurements ...... 44
3.4.1 Line Width and MBH Calculations ...... 48
3.5 Discussion...... 49
3.5.1 The Radius–Luminosity Relationship ...... 49
3.5.2 Comments on Individual Objects ...... 51
Chapter 4 The High-Ionization Component of the BLR in Mrk 335 72
4.1 Background ...... 72
4.2 Observations...... 74
xi 4.2.1 Spectroscopy ...... 74
4.2.2 Photometry ...... 75
4.3 Light Curves and Time Series Analysis ...... 75
4.3.1 LightCurves ...... 75
4.3.2 Time delay measurements ...... 76
4.3.3 Line width measurement and MBH calculations ...... 78
4.4 Discussion...... 79
Chapter 5 The Structure of the BLR: Reconstructed Velocity-Delay Maps ...... 87
5.1 Background ...... 87
5.2 Initial Hβ Velocity-Binned Time Series Analysis ...... 88
5.3 Two-Dimensional MEMECHO Fitting ...... 89
5.3.1 PreparationofSpectra ...... 89
5.3.2 MEMFitting ...... 90
5.3.3 The Driving Continuum Light Curve ...... 92
5.4 Two-Dimensional Velocity-Delay Maps ...... 95
5.4.1 Comments on Individual Objects ...... 96
5.4.2 MEMECHO Parameters and Settings ...... 101
Chapter 6 Stellar Velocity Dispersion Measurements in High- Luminosity Quasar Hosts and Implications for the AGN Black Hole Mass Scale ...... 123
6.1 Background ...... 123
6.2 OBSERVATIONS AND DATA ANALYSIS ...... 124
6.2.1 NIFS/ALTAIRObservations...... 124
6.2.2 DataReduction...... 126
xii 6.2.3 Stellar Velocity Dispersion Measurements ...... 128
6.2.4 Recalculation of AGN Virial Products ...... 132
6.3 ResultsandDiscussion ...... 134
6.3.1 DataQuality ...... 134
6.3.2 The Virial Factor f and the M –σ∗ Relation ...... 136 h i BH 6.4 Summary ...... 140
Chapter 7 Conclusions ...... 159
7.1 Summary ...... 159
7.2 FutureWork...... 163
7.2.1 ModelingtheStructureoftheBLR ...... 163
7.2.2 The RBLR–L Relation and BLR structure in High-Ionization Lines...... 164
7.2.3 The MBH–σ∗ Relation and the AGN MBH Scale ...... 166
Bibliography ...... 167
Appendix A 2010 Reverberation Campaign Light Curves ...... 176
xiii List of Tables
2.1 Continuum and Hβ FluxesforPG2130+099 ...... 34
2.2 PG2130+099LightCurveStatistics...... 35
2.3 PG2130+099ReverberationResults ...... 35
2.4 PG2130+099 Host Galaxy Flux Removal Parameters ...... 36
2.5 Reverberation Results From Kaspi et al. Dataset ...... 37
3.1 2010Reverberation Campaign Object List...... 65
3.2 Spectroscopic and Photometric Observations ...... 66
3.3 SpectralPropertiesofTargets ...... 67
3.4 Continuum and Emission-Line Integration Regions ...... 68
3.5 Continuum Light Curve Statistics ...... 69
3.6 Hβ LightCurveStatistics ...... 70
3.7 Rest-frame Hβ LagMeasurements...... 70
3.8 Mean and RMS Line Widths, Virial Masses and Luminosities. .... 71
4.1 Mrk335LightCurveStatistics ...... 85
4.2 Hβ and He ii λ4686TimeSeriesResults...... 86
5.1 MEMECHOParameters ...... 122
6.1 QuasarProperties...... 151
6.2 GeminiNIFSObservations...... 151
6.3 NIFS Extraction Windows and Measurements ...... 152
xiv 6.4 Most Prominent Stellar Absorption Features ...... 153
6.5 Reverberation Measurements and Updated Virial Products for the FirstPartoftheAGNSample ...... 154
6.6 Reverberation Measurements and Updated Virial Products for the SecondPartoftheAGNSample...... 155
6.7 Reverberation Measurements and Updated Virial Products for the ThirdPartoftheAGNSample ...... 156
6.8 The AGN MBH–σ∗ Sample...... 157
6.8 The AGN MBH–σ∗ Sample...... 158
A.1 V -bandandContinuumFluxes ...... 177
A.1 V -bandandContinuumFluxes ...... 178
A.1 V -bandandContinuumFluxes ...... 179
A.1 V -bandandContinuumFluxes ...... 180
A.1 V -bandandContinuumFluxes ...... 181
A.1 V -bandandContinuumFluxes ...... 182
A.1 V -bandandContinuumFluxes ...... 183
A.1 V -bandandContinuumFluxes ...... 184
A.1 V -bandandContinuumFluxes ...... 185
A.1 V -bandandContinuumFluxes ...... 186
A.1 V -bandandContinuumFluxes ...... 187
A.1 V -bandandContinuumFluxes ...... 188
A.1 V -bandandContinuumFluxes ...... 189
A.1 V -bandandContinuumFluxes ...... 190
A.1 V -bandandContinuumFluxes ...... 191
A.1 V -bandandContinuumFluxes ...... 192
A.2 Hβ Fluxes...... 193
xv A.2 Hβ Fluxes...... 194
A.2 Hβ Fluxes...... 195
A.2 Hβ Fluxes...... 196
A.2 Hβ Fluxes...... 197
A.2 Hβ Fluxes...... 198
xvi List of Figures
2.1 Mean and rms spectra of PG 2130+099 in the observed frame. .... 27
2.2 Continuum and Hβ lightcurves...... 28
2.3 Cross correlation functions for our recent observations of PG2130+099. 29
2.4 Light curves and cross correlation functions from the Kaspi et al. (2000) data...... 30
2.5 Light curves and cross correlation functions using a subset of the Kaspi etal.(2000)data...... 31
2.6 The position of PG2130+099 on the RBLR–L relationship...... 32
2.7 The position of PG2130+099 on the MBH–σ∗ relationship...... 33
3.1 Flux-calibrated mean and rms residual spectra ...... 56
3.2 CompletelightcurvesMrk335andMrk1501...... 57
3.3 Completelightcurvesfor3C120andMrk6...... 58
3.4 CompletelightcurvesforPG2130+099...... 59
3.5 SPEAR predicted light curves for Mrk 335 and Mrk 1501...... 60
3.6 SPEAR predicted light curves for 3C 120 and Mrk 6...... 61
3.7 SPEAR predicted light curves for PG 2130+099 ...... 62
3.8 Cross correlation and log-likelihood functions...... 63
3.9 The Hβ RBLR–L relation...... 64
4.1 MeanandRMSresidualspectraofMrk335 ...... 82
4.2 CompletelightcurvesforMrk335...... 83
xvii 4.3 CCFsforthelightcurves...... 84
5.1 Velocity-binned reverberation lag results...... 105
5.2 Continuum light curves and JAVELIN mean continuum models ... 106
5.3 BestMEMECHOfitstothespectraofMrk335...... 107
5.4 BestMEMECHOfitstothespectraofMrk1501...... 108
5.5 BestMEMECHOfitstothespectraof3C120...... 109
5.6 BestMEMECHOfitstothespectraofMrk6...... 110
5.7 Best MEMECHO fits to the spectra of PG2130+099...... 111
5.8 Best-fit velocity-delay maps over the full rest-frame wavelength range foreachobject...... 112
5.9 Velocity-delay maps for both emission lines seen in Mrk 335...... 113
5.10 Velocity-delay maps for both emission lines seen in Mrk 1501. . . . . 114
5.11 Velocity-delay maps for all three emission lines seen in3C120. . . . . 115
5.12 Velocity-delay maps for all three emission lines seen in PG2130+099. 116
5.13 False-color velocity-delay maps...... 117
5.14 Velocity-delay maps for simple BLR models of Hβ emission...... 118
5.15 One-dimensional delay map for PG 2130+099...... 119
5.16 Velocity-delay maps for 3C 120, varying the A and W parameters. . . 120
5.17 Velocity-delay maps for 3C 120, varying the degree of smoothing and theminimumlagallowed...... 121
6.1 Gemini NIFS: Raw reconstructed images for each object...... 142
6.2 Observed-frame NIFS spectra of PG0026+129 and PG0052+251. . . 143
6.3 Observed-frame NIFS spectra of PG1226+023 and PG1411+442. . . 144
6.4 Observed-frame NIFS spectra of PG1617+175 and PG1700+518. . . 145
6.5 Observed-frame NIFS spectra of Mrk 509 and PG2130+099...... 146
xviii 6.6 Normalized rest-frame spectra of PG1411+442 and PG1617+175. . . 147
6.7 Normalized rest-frame spectra of Mrk 509 and PG2130+099 ..... 148
6.8 The MBH–σ∗ relation...... 149
6.9 The MBH-σ∗ relation in AGNs divided by morphology...... 150
xix Chapter 1
Introduction
1.1. Supermassive Black Holes in Galaxies
It is now widely accepted that most, if not all, massive galaxies host a massive black hole (BH) at their centers. Over the past couple of decades, both observational and analytical work have suggested a physical connection between the formation and growth of galaxies and the growth of their central black holes. Among the lines of evidence for this connection is the existence of correlations between the masses of supermassive black holes (MBH) and various properties of the galaxies in which they reside. Explanations for the observed MBH–galaxy connections have ranged from hierarchical galaxy mergers and AGN feedback to self-regulated BH growth (e.g.,
Silk & Rees 1998; Di Matteo et al. 2005; Hopkins et al. 2009), although there are also arguments that it is simply a consequence of random mergers (e.g., Peng 2007;
Peng 2010; Jahnke & Macci`o2011). The observed MBH–galaxy correlations have the potential to help us constrain the process of galaxy evolution as well as provide a means of measuring the properties of black holes in a wide range of galaxies across cosmologically interesting distances.
1 In order to fully understand these connections and characterize the relationship betwen BHs and their hosts, we require accurate BH mass (MBH) measurements across the observable universe. Direct MBH measurements are made with stellar kinematics and gas dynamics, although these methods require good spatial resolution and are presently only feasible for nearby galaxies. AGNs, discussed below, offer the most robust tracer of the evolution of the BH population over much of the history of the universe.
1.2. Active Galactic Nuclei
AGNs were first observed in the early twentieth century and were noticed because of their strong emission lines. They were first established as a distinct class of objects by Carl Seyfert in 1943. Now termed Seyfert galaxies, they are defined as objects with point-like cores and high-excitation emission lines. Decades later, some Seyfert galaxies were detected as radio sources at high redshifts, and were dubbed “quasi-stellar radio sources”, or quasars. Today, both quasars and Seyfert galaxies are thought to be similar physically, with “quasar” generally referring to higher-luminosity AGNs, while Seyferts are lower-luminosity objects in which the host galaxy is clearly detectable. AGNs have many distinct observational characteristics:
The emission comes from a compact source in the core of the host galaxy. • 2 They have relatively strong continuum emission across the entire • electromagnetic spectum.
They exhibit high continuum variability. •
They often exhibit both broad (FWHM > 1000 km s−1) and narrow (FWHM • < 1000 km s−1) emission lines.
AGN luminosities can exceed the luminosities of the entire galaxies that host them, and thus can be observed out to very high redshifts. Their exact characteristics can vary from object to object, and as a result there are several different classes of
AGNs. For example, objects that do not show the extremely broad emission lines that are present in many others are known as Type 2 AGNs, while AGNs that show both broad and narrow emission lines are classified as Type 1.
The generally accepted powering mechanism in AGNs is the accretion of gas onto a central supermassive BH. Gas falling onto the BH forms a hot accretion disk, which generates the broadband continuum emission we observe. The Doppler- broadened emission lines originate in gas orbiting the BH in the so-called broad line region (BLR), which is photoionized by the accretion disk. The size of the BLR
(RBLR) depends on the AGN luminosity L (see Section 1.3 below), so the exact size of the BLR varies from object to object, and ranges from less than a light-day to hundreds of light-days in size. The narrow emission lines arise from a more extended, lower-density region known as the narrow-line region (NLR) that is on the order
3 of tens of parsecs in size. Surrounding the system is an obscuring torus structure that extends out to kiloparsec scales. The differences between different classes of
AGNs are thought to be caused by different viewing angles of each source. If we are viewing the system from above the torus, nearly along the axis of symmetry, we can see directly into the central region of the AGN, which means we see the broad emission lines typical of Type 1 AGNs. If we are looking at the system through the obscuring torus, the BLR is obscured and we see only the narrow emission lines from the extended NLR and some scattered light, which results in a Type 2 AGN.
AGNs are of particular importance in studies of galaxy formation because we can use the BLR gas to measure MBH in AGNs at much greater distances than is possible using dynamical methods in quiescent galaxies. Under the assumption that the motion of the gas in the BLR is dominated by the gravitational influence of the black hole, we have the virial relation
(fR ∆V 2) M = BLR , (1.1) BH G
where RBLR is the average distance of the emitting gas in the BLR from the BH and is usually either determined using reverberation mapping (e.g., Peterson et al. 2004) or estimated using the relation between RBLR and the AGN luminosity (e.g., Bentz et al. 2009a; Bentz et al. 2013). The velocity dispersion (∆V ) is deduced from the width of the emission line, G is the gravitational constant, and f is a dimensionless factor that accounts for the unknown geometry and orientation of the BLR and
4 is different for each individual AGN. With current technology, we are unable to directly observe the structure of the BLR, as it is unresolvable even with the largest telescopes, so the true value of f for each object is unknown. This has contributed significantly to the uncertainties in MBH measurements using BLR emission lines.
Reverberation mapping techniques, described below, provide a means for measuring
RBLR and determining the structure of the BLR itself.
1.3. Reverberation Mapping
Reverberation mapping uses observations of continuum and emission-line variability to probe the structure of the BLR (Blandford & McKee 1982; Peterson
1993). We measure the radius of the BLR by monitoring the thermal continuum flux coming from near the BH and the total flux in a nearby broad emission line (e.g.,
Hβ) from the BLR and measuring the time τ it takes for light variations from the central source to propagate out to the BLR. This time delay between the variations is assumed to be the mean light-travel time across the BLR, and thus yields an estimate of RBLR (RBLR = cτ, where c is the speed of light).
Reverberation mapping has been extensively used to estimate the physical size of the BLR and the mass of central black holes in AGNs, and to date has been applied to measure average BLR radii and MBH in around 50 AGNs (e.g.,
Peterson et al. 2004; Bentz et al. 2009c; Denney et al. 2010). However, light curve
5 quality, in terms of sampling density, duration, and precision flux measurements, is a very important factor in reverberation measurements. In particular, light curves that are too short in duration or inadequately sampled can result in incorrect lag measurements (e.g., Perez et al. 1992; Welsh 1999; Grier et al. 2008). Since the early
1990s, the view of what constitutes “adequately-sampled” has changed dramatically.
We now know that some of the early measurements need to be redone, as their sampling rates are low enough that we have serious doubts about their suitability in recovering BLR radii. Improving time delay measurements is the motivation behind much of my work.
A useful relationship that has emerged from reverberation mapping is the correlation between RBLR and the optical luminosity of the AGN (e.g. Kaspi et al.
2000, 2005; Bentz et al. 2009a; Bentz et al. 2013), known as the RBLR–L relation.
This relation is critical because it allows us to estimate MBH in AGN from a single spectrum — we get RBLR from the AGN luminosity, the velocity dispersion of the BLR is determined by the emission-line width, and the quiescient MBH–σ∗ relationship, discussed below, provides the calibration of the reverberation-based mass scale. The RBLR–L relation is predicted by photoionization theory (e.g.,
Davidson 1972; Davidson & Netzer 1979) and is remarkably tight. Using the RBLR–L relation, we can obtain both velocity and RBLR estimates from a single calibrated spectrum. As such, this relation has been used to calculate MBH in large samples of
AGNs (e.g., Shen et al. 2008). These large samples can be used to investigate the
6 evolution of the BH mass function (e.g., Greene & Ho 2007; Vestergaard et al. 2008;
Vestergaard & Osmer 2009; Kelly et al. 2010), the growth of BHs compared to their hosts, the Eddington ratios of quasars (e.g., Kollmeier et al. 2006; Kelly et al. 2010), and even the dependence of accretion disk sizes on BH mass (Morgan et al. 2010).
Because of its broad influence on our understanding of AGNs, it is important that our characterizations of the RBLR–L relations for various emission lines are accurate.
1.4. Velocity-Resolved Reverberation Mapping
Most reverberation studies have been limited to measuring the mean time delay τ for various emission lines (e.g., Peterson et al. 2004; Denney et al. 2010; h i Bentz et al. 2010a; Grier et al. 2012b). This allows us to estimate the mean radius of the BLR and estimate MBH, but it reveals very little information about the detailed structure of the BLR. To learn about the BLR structure, we must rely on velocity-resolved reverberation techniques or the microlensing of gravitationally lensed quasars (e.g., Guerras et al. 2013). Mathematically, the variations in the BLR emission-line flux, ∆L(V, t), can be described as a convolution of the continuum
flux variations, ∆C(t), with the “transfer function”, Ψ(V, τ) (Blandford & McKee
1982). The transfer function depends on both the temporal lag τ between the
7 line and continuum emission and the line-of-sight velocity V of the BLR gas. The relationship is expressed mathematically as
∞ ∆L(V, t)= Ψ(V, τ)∆C(t τ)dτ. (1.2) Z0 −
A main goal of reverberation mapping is to recover Ψ(V, τ), also called the
“velocity-delay map”, which describes how the continuum flux variations give rise to
BLR flux variations across entire emission lines, and therefore contains information about the BLR geometry and kinematics. For example, a Keplerian disk produces a velocity-symmetric structure in Ψ(V, τ), with a wider/narrower range of velocities at smaller/larger delays. In contrast, radial flows give rise to asymmetric velocity structure, the signature of infall/outflow being smaller delays on the red/blue side of the velocity profile. Only very recent reverberation mapping efforts have been successful in obtaining detailed information about the actual structure of the BLR in some objects (e.g., Bentz et al. 2010b; Brewer et al. 2011; Pancoast et al. 2012;
Grier et al. 2013).
1.5. The AGN MBH–σ Relation ∗
Among the many lines of evidence pointing to a connection between BHs and their host galaxies are several different correlations observed between MBH and properties of the hosts. One of the best-studied correlations between MBH and host
8 galaxy properties is that between MBH and the stellar velocity dispersion of the host bulge σ∗ (commonly known as the MBH–σ∗ relation). The MBH–σ∗ relation was
first predicted by Silk & Rees (1998) and Fabian (1999) and has been explained by various analytic models (e.g., King 2003; King 2005; Murray et al. 2005) as well as recovered in numerical simulations of evolving and interacting galaxies (e.g.,
Di Matteo et al. 2005; Di Matteo et al. 2008). The MBH–σ∗ relation is observed in both quiescent (Ferrarese & Merritt 2000; Gebhardt et al. 2000; Tremaine et al.
2002; G¨ultekin et al. 2009; McConnell et al. 2011; McConnell & Ma 2013) and active galaxies (Gebhardt et al. 2000b; Ferrarese et al. 2001; Nelson et al. 2004; Onken et al. 2004; Dasyra et al. 2007; Woo et al. 2010; Graham et al. 2011; Park et al.
2012). The MBH–σ∗ relation is extremely useful because it can be used to infer MBH in large samples of galaxies. This allows for the exploration of the BH mass function on much larger scales (e.g., Yu & Tremaine 2002) and thus helps investigate the role of BHs in galaxy formation and evolution processes.
Beyond its utility in testing models of galaxy evolution, the quiescent MBH–σ∗ relation has also been used to calibrate MBH measurements in AGNs. Recent reverberation mapping efforts have begun to reveal more information about the actual structure of the BLR and the value of f in some objects (e.g., Bentz et al.
2010b; Brewer et al. 2011; Pancoast et al. 2012; Grier et al. 2013). However, limited data for most AGNs requires the use of an average virial factor f to estimate M . h i BH
Currently, f is calculated with the assumption that AGNs follow the same M –σ∗ h i BH
9 relation as quiescent galaxies (Onken et al. 2004; Woo et al. 2010; Graham et al.
2011; Park et al. 2012). Most estimates of f are somewhat larger than 5; Onken h i ∼ et al. (2004) find f = 5.5 1.8, Woo et al. (2010) find f = 5.2 1.2, and more h i ± h i ± recently, analysis by Park et al. (2012) yields f = 5.1 0.8. However, Graham h i ± +0.7 et al. (2011) obtain a lower value, f = 2.8− . h i 0.5
Park et al. (2012) found that the disparity between the recently reported slopes and virial factors of Graham et al. (2011) and the others (Onken et al. 2004;
Woo et al. 2010; Park et al. 2012) is caused mostly by the type of regression used in the fit (whether MBH is considered the independent or dependent variable).
Another difference arises when different galaxy samples are used to determine these two quantities. Some recent studies report a morphological dependence in the quiescent MBH–σ∗ relation, such that early-type galaxies exhibit a steeper slope
(e.g., McConnell & Ma 2013). Graham & Scott (2013) also find a morphological dependence in the MBH–σ∗ relation. They note that galaxies whose core structure is thought to be created through dry merger processes follow a different slope on the relation than galaxies thought to be built through gas-rich merger processes. The idea of a non-universal MBH–σ∗ relation has been supported by theoretical work as well (e.g., King 2010; Zubovas & King 2012), which has also suggested that the relation may depend on environment.
Deviations from a single MBH–σ∗ relation have also been claimed in AGNs, and there has been some question as to whether or not objects at the high-mass/high-σ∗
10 end of the relation follow a different slope (e.g., Dasyra et al. 2007; Watson et al.
8 2008). For example, four out of the six objects with MBH above 10 M⊙ included in the study of Watson et al. (2008) lie significantly above the relation. The appearance of outliers could be due to systematic errors in σ∗ or MBH measurements, or simply a fluke due to small number statistics. Alternatively, Lauer et al. (2007) suggest that offsets at the high-mass end may be due to a selection bias. Specifically, when a sample is selected based on AGN properties, one is more likely to find a high-mass
BH in a lower-mass galaxy (based on a BH–host galaxy correlation) because high mass galaxies are rare and there is intrinsic scatter in BH–host galaxy correlations.
1.6. Scope of this Dissertation
In this dissertation, I will discuss the results of a major effort to improve the AGN BH mass scale through reverberation mapping experiments aimed at uncovering the structure of the BLR, as well as efforts to improve the MBH–σ∗ relation and thus the calibration of the AGN mass scale as a whole.
In Chapter 2, I discuss the results of a reverberation campaign carried out in
2007, aimed at the quasar PG 2130+099. This chapter has been published as “The
Mass of the Black Hole in the Quasar PG 2130+099” by C. J. Grier et al., 2008, The
Astrophysical Journal, 688, 837.
11 In Chapter 3, I discuss our massive reverberation campaign carried out in the fall of 2010 with the goal of measuring the Hβ time lag in several nearby AGNs. I discuss the data collection and analysis, as well as initial reverberation results. This chapter has been published as “Reverberation Mapping Results for Five Seyfert 1
Galaxies” by C. J. Grier et al., 2012, The Astrophysical Journal, 755, 60.
In Chapter 4, I discuss one special object observed in this campaign, Mrk 335, for which we report the first robust reverberation measurement of a high-ionization emission line. This has been published as “A Reverberation Lag for the High-
Ionization Component of the Broad-Line Region in the Narrow-Line Seyfert 1 Mrk
335” by C. J. Grier et al., 2012, The Astrophysical Journal Letters, 744, 4.
In Chapter 5, I discuss the velocity-resolved results for the campaign and the velocity-delay maps recovered for four objects, which has been published as “The
Structure of the Broad Line Region in Active Galactic Nuclei. I. Reconstructed
Velocity-Delay Maps” by C. J. Grier et al., 2013, The Astrophysical Journal, 764,
47.
In Chapter 6, I discuss the measurement of stellar velocity dispersions in the host galaxies of high-luminosity AGNs and a recalculation of the virial factor f. The goal of this last project is to improve the overall calibration of MBH measurements in AGNs, which currently relies on the MBH–σ∗ relation. This version of this work has been submitted for publication by the Astrophysical Journal as “Stellar Velocity
12 Dispersion Measurements in High-Luminosity Quasar Hosts and Implications for the AGN Black Hole Mass Scale” by C. J. Grier et al., 2013. Finally, in Chapter 7,
I summarize the results presented in this dissertation and discuss potential future directions for research.
All of the publications have been reformatted to comply with Graduate School dissertation formatting rules and have been edited to improve the flow of the dissertation.
13 Chapter 2
The Mass of the Black Hole in the Quasar PG 2130+099
2.1. Background
PG 2130+099, a previously reverberation-mapped quasar, has long been a source of curiosity because it is an outlier in both the MBH–σ∗ and RBLR–L relations.
Initial reverberation measurements reported by Kaspi et al. (2000) and Peterson et al. (2004) found Hβ lags of about 180 days. These measurements yield a black
8 44 hole mass M upwards of 10 M⊙. At luminosity λL (5100 A)=(2˚ .24 0.27) 10 BH λ ± × −1 erg s , this places PG 2130+099 well above the RBLR–L relationship (Bentz et al.
2009a). Dasyra et al. (2007) also note that PG2130+099 falls above the MBH–σ∗ relationship. While it is possible that PG 2130+099 is a real, physical outlier on these relations, we must consider the possibility of measurement errors in the BLR radius. Our suspicions of errors are fueled by two other factors. First, the optical spectrum of PG 2130+099 is similar to that of narrow-line Seyfert 1 (NLS1) galaxies, which are widely supposed to be AGN with high accretion rates relative to the
Eddington rate (see Komossa 2008 for a recent review). However, the accretion
14 rate derived from this mass and luminosity (under common assumptions for all reverberation-mapped AGN as described by Collin et al. 2006 is quite low compared to NLS1s, again suggesting that MBH and therefore τ are overestimated. Second, as we discuss further below, a lag of approximately half a year on an equatorial source means that fine-scale structure in the two light curves will not match up in detail, and cross-correlation becomes very sensitive to long-term secular variations that may or may not be an actual reverberation signal.
For these reasons, we decided to undertake a new reverberation campaign to remeasure the Hβ lag for PG 2130+099. In this chapter, we present a new Hβ lag determination for PG 2130+099 from this campaign. We also present a re-analysis of the earlier dataset that suggests that the true lag is consistent with our new lag value and investigate possible sources of error in the previous analysis.
2.2. Observations and Data Analysis
2.2.1. Observations
The data were obtained as a part of queue-scheduled program during the
SDSS-II Supernova Survey follow-up campaign (Frieman, J. A. et al. (2008)). We obtained spectra of PG 2130+099 with the Boller and Chivens CCD Spectrograph on the MDM 2.4m telescope for 21 different nights from 2007 September through
15 2007 December. We used a 150 grooves/mm grating which yields a dispersion of 3.29
A˚ pixel−1 and covers the spectral range 4000–7500 A.˚ The slit width was set to 3′′.0 projected on the sky and the spectral resolution was 15.2 A.˚ We used an extraction aperture that corresponds to 3′′.0 7′′.0. ×
2.2.2. Light Curves
The reduced spectra were flux-calibrated using the flux of the [O iii] λ5007 emission line in a reference spectrum created using a selection of 9 nights with the best observing conditions. Using a χ2 goodness-of-fit estimator method (van
Groningen & Wanders 1992), each individual spectrum was scaled to the reference spectrum. For three epochs, no reasonable fit could be obtained, hence we scaled these spectra by hand to match the [O iii] λ5007 flux of the reference spectrum, which was 1.36 10−13 erg s−1 cm−2 in the observed frame (z = 0.06298). We then × removed the narrow-line components of the Hβ and [O iii] λλ 4959, 5007 lines using relative line strengths given by Peterson et al. (2004). Figure 2.1 shows the mean and rms spectra of PG2130+099, created from the entire set of 21 flux-calibrated observations.
To measure the continuum and line fluxes, we fit a line to the continuum between the regions 5050–5070 A˚ and 5405–5435 A˚ in the observed frame. We measured the
Hβ flux by integrating the flux above the continuum from 5070–5285 A,˚ and the
16 optical continuum flux was taken to be the average flux in the range 5405–5435 A.˚
Errors were estimated based on differences between observations that were close together in time. The observations from 2007 September 27 (JD2454371) had an extremely low signal-to-noise ratio (S/N) compared to the rest of the population; following Peterson et al. (1998a), we assigned fractional errors of 1/(S/N) for this night in both the continuum and line flux. The resulting continuum and Hβ light curves are given in Table 2.1 and are shown in Figure 2.2. The properties of the final light curves used in our time series analysis are given in Table 2.2. Column (1) gives the spectral feature and Column (2) gives the number of points in the individual light curves. Columns (3) and (4) give the average and median time spacing between observations, respectively. Column (5) gives the mean flux of the feature in the observed frame, and Column (6) gives the mean fractional error that is computed based on observations that are closely spaced. Column (7) gives the excess variance, defined by
√σ2 δ2 F = − (2.1) var f h i where σ2 is the flux variance of the observations, δ2 is the mean square uncertainty, and f is the mean observed flux. Column (8) is the ratio of the maximum to h i minimum flux in each light curve.
We note in passing that we attempted to determine light curves for the other prominent emission lines in our spectra, Hα and Hγ. Unfortunately, the fidelity
17 of our flux calibration decreases away from the [O iii] λ5007 line (which we believe accounts for the increasing strength of the rms spectrum longward of 5800 A˚ in ∼ Figure 2.1) and, combined with the low amplitude of variability, renders the other light curves unreliable. Nevertheless, time series analysis yields results that are at least consistent with the Hβ results, though with much larger uncertainties.
2.3. Time Series Analysis
2.3.1. Time delay measurements
To measure the lag between the optical continuum and Hβ emission-line variations, we cross-correlated the Hβ light curve with the continuum light curve using the interpolation method originally described by Gaskell & Sparke (1986) and Gaskell & Peterson (1987) and subsequently modified by White & Peterson
(1994) and Peterson et al. (1998b, 2004). The method works as follows: because the data are not evenly spaced in time, we interpolate between points to obtain an evenly-sampled light curve. The linear correlation coefficient r is then calculated using pairs of points, one from each light curve, separated by a given time lag. When r is calculated for a range of time lag values, the cross correlation function (CCF) is obtained, which consists of the value of r for each time lag value. We interpolated between the points on a 0.5 day timescale to obtain the CCF for our dataset, which is shown in Figure 2.3.
18 To determine the most probable delay and its uncertainty, we employed the flux randomization/random subset sampling Monte Carlo method described by Peterson et al. (1998b). The “random subset sampling” method takes a light curve of N points and samples it N times without regard to whether or not any given point has been selected already. The flux uncertainty of a data point selected n times is correspondingly reduced by a factor n−1/2 (Welsh 1999; Peterson et al. 2004). The
flux value of each point is then altered by a random Gaussian deviate based on the uncertainty assigned to the point; this is known as “flux randomization”. A CCF was calculated for each altered light curve using interpolation as before. Using 2000 iterations of this process, we obtain a distribution of τpeak, measured from the CCF of each Monte Carlo realization and defined by the location of the peak value of r.
We also calculate and obtain a distribution of the lag values τcent that represents the centroid of each CCF using the points surrounding the peak, based on all lag values with r 0.8r . We then adopt the mean values of both the centroid and ≥ max peak τ and τ for our analysis. We estimate the uncertainties in τ and h centi h peaki cent
τpeak such that 15.87% of the realizations fall below the mean minus the lower error and 15.87% of the realizations fall above the mean plus the upper error. Our final
+4.4 +5.6 lag values and errors are τcent = 22.9−4.3 days and τpeak = 22.2−5.2 days in the rest frame of PG2130+099. It is important to note that the CCF is heavily dependent on the timespan of the light curve relative to the actual delay, and if the data timespan is short with respect to the lag, the CCF will be biased towards short lag
19 measurements (Welsh 1999). We note that because our data span just over three months, our CCF is not capable of producing a lag measurement as large as previous measurements and is subject to this bias. However, if the actual lag is shorter as the evidence presented in this work suggests, our data span a sufficiently long period to identify the true delay.
As a check on our time series analysis, we also employed an alternative method known as the z-transformed discrete correlation function (ZDCF), as described by
Alexander (1997). This method is a modification of the discrete correlation function
(DCF) described by Edelson & Krolik (1988). The DCF is obtained by correlating data points from the continuum with points from the Hβ light curve by binning the data in time rather than interpolating between points. The ZDCF similarly uses bins rather than interpolation, but bins the data by equal population rather than by equal spacing in time and applies Fisher’s z-transform to the cross correlation coefficients. Discrete correlation functions have the advantage that they only use real data points, and are therefore less likely to find spurious correlations in data where there are large time gaps. However, datasets with few points tend to yield lag values with large uncertainties; hence this method is primarily useful as a check on the interpolation method when the data are undersampled. Figure 2.3 shows the computed ZDCF for our PG 2130+099 dataset. It should be noted that the z-transform requires a minimum of 11 points per bin to be statistically significant
(Alexander 1997); because our light curves contained only 21 points, we were unable
20 to obtain a well-sampled ZDCF with a minimum of 11 points per bin. We used a minimum of 8 points per bin and do not obtain an independent lag measurement, but the ZDCF is very consistent with our calculated CCF, indicating that the interpolation results are credible.
2.3.2. Line width measurement and mass calculations
We use the second moment of the line profile, σline (Fromerth & Melia 2000;
Peterson et al. 2004) to characterize the width of the Hβ line. To determine the best value of σline and its uncertainties, we use Monte Carlo simulations similar to those used in determining the time lag (see Peterson et al. 2004). We measure the line widths from both the mean and rms spectra created in the simulations from
N randomly chosen spectra and correct them for the spectrograph resolution. We obtain a distribution of each line width from multiple realizations, from which we take the mean value for our measure of σline or FWHM and the standard deviation as the measurement uncertainty. The measured line widths and uncertainties are given in Table 2.3.
Assuming that the motion of the Hβ-emitting gas is dominated by gravity, the
2 relation between MBH, line width, and time delay is MBH = fcτ∆V /G, where τ is the emission-line time delay, ∆V is the emission-line width, and f is a dimensionless factor that is characterized by the geometry and kinematics of the BLR. Onken et al.
21 (2004) calculated an average value of f by assuming that AGN follow the same h i
MBH–σ∗ relationship as quiescent galaxies. By normalizing the reverberation black hole masses to the relation for quiescent galaxies, they obtain f = 5.5 when the h i 1 line width is characterized by the line dispersion σline of the rms spectrum. Using
7 our values of τ and σ , we compute M = (3.8 1.5) 10 M⊙. cent line BH ± ×
We also compute the average 5100 A˚ luminosity of our sample to see where our new RBLR value places PG 2130+099 on the RBLR–L relationship. Following
Bentz et al. (2009a, 2013), we correct our luminosity for host-galaxy contamination.
−1 −1 Assuming H0 = 70 km s Mpc , Ωm = 0.30 and ΩΛ = 0.70, we calculate log λL (5100 A)=˚ 44.40 0.02 for our recent measurements of PG 2130+099. The λ ± relevant measurements and computed quantities are given in Table 2.4.
2.4. Discussion
Analysis of previous datasets of PG 2130+099 yields lags greater than 150 days
8 and MBH values above 10 M⊙, in both cases approximately an order of magnitude larger than our new value. To investigate the source of these discrepancies, we closely scrutinized the previous dataset, which consists of data obtained at Steward
Observatory and Wise Observatory (Kaspi et al. 2000). The 5100 A˚ continuum and
1 Differences between the use of σline and the FWHM in calculating MBH are discussed by Collin et al. (2006).
22 emission-line light curves, along with their respective CCF and ZDCFs, are shown in Figure 2.4. We first ran the full light curve through the time series analysis as described in Section 2.3.1 to confirm the previous results, and successfully reproduced a lag measurement of 168 days, in agreement with Kaspi et al. (2000) ∼ and Peterson et al. (2004). However, visual inspection of the light curves suggests that these values were in error; we were able to identify several features that are present in the continuum, Hβ and Hα light curves that are quite visibly lagging on timescales shorter than 50 days, as we show below.
It is clear that the Kaspi et al. results are dominated by the data spanning the years 1993-1995, as this period contains the most significant variability as well as time sampling that is sufficient to resolve features in the light curves. We show the light curves for all spectral features measured by Kaspi et al. during this time period in Figure 2.5. Over this period, we can match behavior in the continuum with the behavior of the emission lines, most noticeably in the data from 1995. The most prominent features in the light curves are the maximum in the continuum at the end of the 1994 series and the maxima in the lines at the beginning of the 1995 series, which, based on the continuum variations, one would expect to be lower relative to the fluxes in 1994. However, inspection of the relative flux levels of the lines and continuum in the 1994 and 1995 data reveals that the equivalent width of the emission lines changes during this timespan, which results in a rise in emission-line
flux apparently unrelated to reverberation. The maximum in the continuum and
23 those in the emission lines do not correspond to the same features— the maxima in the emission line light curves in 1995 correspond to the similar feature in the continuum at the beginning of 1995. Because there are no observations during the six months or so between these features, the CCF and ZDCF lock onto the two unrelated maxima that are separated by 196 days and yield lags of 200 days over ∼ this three-year span. The emission lines likely continued to increase in flux during the time period for which there were no observations.
Inspection of the light curve segments in Figure 2.5 reveals similar structures in each of them within a given year. To quantify the time delays between the continuum and lines on short timescales, we cross-correlated each individual year, this time using only flux randomization in the Monte Carlo realizations, as there were too few points in each year to use random subset sampling, so the uncertainties are underestimated. The resulting lags are given in Table 2.5. Again we must consider that the CCF cannot produce a lag that is greater than the duration of observations, so the individual CCFs for these three years are limited to short delays. However, the presence of the 1995 feature in both the continuum and emission-line light curves is unmistakable and it is extremely unlikely that it lags by a value exceeding the sensitivity limit of the CCF. From this we surmise that the discrepancies in measured lag values are mostly a result of large time gaps in the data and/or underestimation of the error bars in the Kaspi et al. data.
24 Welsh (1999), and even earlier, Perez et al. (1992), pointed out that emission-line lags can be severely underestimated with light curves that are too short in duration, particularly if the BLR is extended: certainly our 98-day campaign is insensitive to lags as long as 180 days. Is it possible that the original lag determination of Kaspi ∼ et al. (2000) and Peterson et al. (2004) is correct (or more nearly so) and we have been fooled by reliance on light curves that are too short? We think not, based on (1) the reasonable match between details in the continuum and emission-line light curves for four different observing seasons (1993, 1994, 1995, and 2007), (2) the improved agreement with the RBLR-L relationship with the smaller lag (demonstrated in
Figure 2.6), and (3) the improved agreement with the MBH-σ∗ relationship with the smaller lag (Figure 2.7). It is also worth pointing out the difficulty of accurately measuring a 180 day lag, particularly in the case of equatorial sources which ∼ have a relatively short observing season. The short observing season, typically
6-7 months, means that there are very few emission-line observations that can be matched directly with continuum points: the observed emission-line fluxes represent a response to continuum variations that occurred when the AGN was too close to the Sun to observe.
Welsh (1999) has also pointed out the value in “detrending” the light curves— removing long-term trends by fitting the light curves with a low-order function can reduce the bias toward underestimating lags. In this particular case, we find that detrending has almost no effect: in particular, the highest points in the continuum
25 (in late 1994) and the highest points in the line (early in the 1995 observing season) remain so after detrending, and both the interpolation CCF and ZDCF weight these points heavily. It is also interesting to note that the peak in the ZDCF agrees with the peak of the interpolation CCF (Figure 2.4), which demonstrates that the
180 day lag is not simply ascribable to interpolation across the gap between the ∼ 1994 and 1995 observing seasons.
The data from our 2007 campaign are not ideal: the amplitude of variability was low, the time sampling was adequate, but only barely, and the duration of the campaign was short enough that we lack sensitivity to lags of 50 days or ∼ longer. But the preponderance of evidence at this point argues that our smaller lag measurement is more likely to be correct than the previous determination. Certainly better-sampled light curves of longer duration would yield a more definitive result.
26 Fig. 2.1.— The flux-calibrated mean and rms spectra of PG 2130+099 in the observed frame (z = 0.06298). The flux density is in units of 10−15 erg s−1 cm−2 A˚−1.
27 5.2
5
4.8
4.6
4.4
4.2
5.2
5
4.8
4.6 54360 54380 54400 54420 54440
Fig. 2.2.— Continuum (upper panel) and Hβ (lower panel) light curves that were used in the time series analysis. Continuum flux densities are in units of 10−15 erg s−1 cm−2 A˚−1. Emission-line flux densities are in units of 10−13 erg s−1 cm−2. All fluxes are in the observed frame.
28 Fig. 2.3.— The CCF (solid line) and ZDCF (filled circles) from the time series analysis of our recent observations of PG 2130+099.
29 Fig. 2.4.— Light curves and cross correlation functions for different spectral features of the Kaspi et al. (2000) data. The left column shows the light curves of PG2130+099 from Kaspi et al. (2000). The 5100 A˚ continuum light curve is shown in the top left panel and the Hα, Hβ, and Hγ light curves are displayed below. The right column shows the CCF and corresponding ZDCF for each spectral feature, with the top right panel showing the auto-correlation function (ACF) of the continuum. The continuum flux density is given in units of 10−15 erg s−1 cm−2 A˚−1 and the Hβ flux density is in units of 10−13 erg s−1 cm−2.
30 Fig. 2.5.— Light curves and cross correlation functions for different spectral features using a subset of the Kaspi et al. (2000) data. The left column shows a small section of the light curve for each spectral feature. The right column panels show the CCF (solid line) and ZDCF (filled circles) for each feature. The top right panel shows the ACF of the continuum, and the other three panels below show the correlation functions resulting from correlation of each respective emission line with the continuum light curve. The fluxes are given in the same units as in Figure 2.4.
31 Fig. 2.6.— The position of PG 2130+099 on the most recent RBLR–L relationship from Bentz et al. (2009), denoted by the solid line. The starred circle represents the position of PG 2130+099 with a lag of 158.1 days, as given by Peterson et al. (2004). The filled square represents the measurement of 22.9 days from our recent dataset. The open squares represent other objects from Bentz et al.
32 Fig. 2.7.— The position of PG2130+099 on the MBH–σ∗ relationship (from Tremaine et al. 2002), denoted by the solid line. The open squares show data points based on Onken et al. (2004) and Nelson et al. (2004) and the open circle is PG1426+015 from Watson et al. (2008). The starred circle and filled square represent the position of PG 2130+099 using the same data sets as in Figure 2.6.
33 JD Fλ (5100 A)˚ Hβλ4861 ( 2400000) (10−15 erg s−1 cm−2 A˚−1) (10−13 erg s−1 cm−2) −
54352.70 4.36 0.08 4.94 0.07 ± ± 54353.70 4.23 0.07 4.98 0.07 ± ± 54354.72 4.38 0.08 4.88 0.07 ± ± 54371.76 4.56 0.15 4.91 0.16 ± ± 54373.65 4.53 0.08 4.82 0.07 ± ± 54382.65 4.53 0.08 4.73 0.07 ± ± 54383.63 4.74 0.08 4.83 0.07 ± ± 54384.64 4.74 0.08 4.88 0.07 ± ± 54392.62 4.93 0.09 4.92 0.07 ± ± 54393.59 4.89 0.08 5.00 0.07 ± ± 54394.57 4.93 0.08 4.95 0.07 ± ± 54395.59 5.14 0.09 5.02 0.07 ± ± 54400.63 4.86 0.08 5.01 0.07 ± ± 54401.63 4.76 0.08 5.01 0.07 ± ± 54413.57 4.40 0.08 5.27 0.08 ± ± 54414.58 4.42 0.08 5.07 0.07 ± ± 54415.57 4.55 0.08 5.21 0.08 ± ± 54416.56 4.47 0.08 5.14 0.07 ± ± 54427.57 4.78 0.08 5.17 0.07 ± ± 54449.56 4.55 0.08 4.84 0.07 ± ± 54450.57 4.69 0.08 5.03 0.07 ± ±
Table 2.1. Continuum and Hβ Fluxes for PG2130+099
34 Sampling Mean Time Interval (days) Mean Fractional Series N T T Flux1 Error F R h i median var max (1) (2) (3) (4) (5) (6) (7) (8)
5100 A˚ 21 4.9 1.0 4.64 0.23 0.018 0.047 1.22 0.03 ± ± Hβ 21 4.9 1.0 4.98 0.14 0.015 0.023 1.14 0.02 ± ±
1Continuum and emission-line fluxes are given in the same units as Table 2.1.
Table 2.2. PG 2130+099 Light Curve Statistics
Parameter Value (1) (2)
a +4.7 τcent 22.9 −4.6 days a +5.6 τpeak 22.2 −5.2 days σ (mean) 1807 4 km s−1 line ± FWHM (mean) 2807 4 km s−1 ± σ (rms) 1246 222 km s−1 line ± FWHM (rms) 2063 720 km s−1 ± 7 M (3.8 1.5) 10 M⊙ BH ± ×
aValues are given in the rest frame of the object.
Table 2.3. PG 2130+099 Reverberation Results
35 Parameter Value (1) (2)
Position angle of slit 0.0◦ Aperture size 3′′.0 7′′.0 × F (5100 A)˚ a (0.405 0.037) 10−15 erg s−1 cm−2 A˚−1 galaxy ± × log λL (5100 A)˚ b (44.40 0.02) λ ±
aGalaxy flux is in the observed frame, from Bentz et al. (2009). b5100 A˚ rest-frame luminosity in erg s−1, with host galaxy subtracted and corrected for extinction, following Bentz et al. (2006; 2009).
Table 2.4. PG 2130+099 Host Galaxy Flux Removal Parameters
36 Data Spectral τcent τpeak Subset Feature (days) (days)
+32.1 +20.0 Entire Data Set Hα 215.2 −27.8 217.4 −20.0 +36.7 +60.0 Entire Data Set Hβ 168.2 −21.4 177.4 −100.0 +37.9 +40.0 Entire Data Set Hγ 197.6 −28.5 208.3 −80.0 +4.8 +10.0 1993-1995 Hα 199.1 −0.7 203.0 −10.0 +5.2 +0.0 1993-1995 Hβ 203.5 −17.4 202.9 −10.0 +35.1 +21.0 1993-1995 Hγ 195.6 −27.6 219.5 −93.0 +0.6 +0.0 1993 Hα 15.6 −1.4 19.1 −3.0 +1.5 +3.0 1993 Hβ 12.2 −2.5 13.1 −2.0 +0.9 +0.0 1994 Hα 15.2 −1.6 16.0 −0.0 +1.8 +0.0 1994 Hβ 13.7 −0.8 16.0 −1.0 +5.3 +51.0 1994 Hγ 10.0 −3.4 12.0 −9.0 +4.7 +0.0 1995 Hα 36.6 −3.3 44.0 −10.0 +4.0 +8.0 1995 Hβ 46.5 −2.9 44.0 −0.0 +7.7 +7.7 1995 Hγ 67.4 −8.5 67.0 −8.5
Table 2.5. Reverberation Results From Kaspi et al. Dataset
37 Chapter 3
Reverberation Mapping Results for Five Seyfert 1 Galaxies
3.1. Background
In a continuing effort to improve the database of reverberation-mapped objects, we carried out a massive reverberation mapping program at multiple institutions beginning in 2010 August and running until 2011 January. The main goals of our program were (1) to re-observe old objects lacking well-sampled light curves, (2) to expand the reverberation-mapped sample by observing new objects, (3) to obtain velocity-delay maps for several of the targets, and (4) if possible, to measure a reverberation lag in the high-ionization He ii λ4686 emission line in a narrow-line
Seyfert 1 galaxy (Mrk 335 in this case, with results discussed in Chapter 4 and published in Grier et al. 2012a). We limited our target list to galaxies with expected time lags that were short enough to allow successful measurements during our four-month long campaign. Our final target list included eight objects, and we succeeded in measuring lags for six. Two objects, NGC 4151 and NGC 7603, were dropped due to weather-related time losses. In this chapter, we present Hβ lag
38 measurements for five of the six remaining objects and the resulting black hole masses. These five targets and their basic properties are listed in Table 3.1.
3.2. Observations
In general, we follow the observational and data reduction practices of Denney et al. (2010) for the spectroscopic observations. Our data analysis methods follow those of Peterson et al. (2004). A brief summary and any deviations from these methodologies are discussed below. When needed, we adopt a cosmological model
−1 −1 with Ωm =0.3, ΩΛ =0.70, and H0 = 70 km sec Mpc .
3.2.1. Spectroscopy
The majority of the spectra were obtained using the MDM Observatory
1.3m McGraw-Hill telescope on Kitt Peak. We used the Boller and Chivens CCD spectrograph to obtain spectra over the course of 120 nights from 2010 August 31 to
December 28. We used the 350 mm−1 grating to obtain a dispersion of 1.33 A˚ pixel−1.
We set the grating for a central wavelength of 5150 A,˚ which resulted in spectral coverage from roughly 4400 A˚ to 5850 A.˚ The slit was oriented north-south (Position
Angle=0) and set to a width of 5′′.0, which resulted in a spectral resolution of 7.9
A.˚ We used an extraction window of 12′′.75 along the slit. We also obtained spectra during this time period using the 2.6m Shajn telescope at the Crimean Astrophysical
39 Observatory (CrAO). These data were acquired with the Nasmith spectrograph and SPEC-10 CCD. A 3′′.0 slit was used at a position angle of 90◦, and we used an extraction window of 13′′.0. Because of the large slit size used, there should be no effect on the AGN light due to the change in position angle between the MDM and
CrAO spectra. However, this will affect the amount of host galaxy light received through the slit. The spectral coverage in the CrAO data was from approximately
3900 A˚ to 6100 A,˚ with a dispersion of 1.0 A˚ pixel−1. Table 3.2 lists the number of spectroscopic observations and time coverage at each telescope for our sample.
The reduced spectra were calibrated onto an absolute flux scale by assuming that the [O iii] λ5007 narrow-line flux is constant. The reference spectra for this calibration were created by averaging spectra taken on photometric nights for each source. We scaled these reference spectra to the absolute flux of the [O iii] λ5007 line for each object (listed in Column 3 of Table 3.3) to create an absolute flux-calibrated reference spectrum for each object. We confirmed that the [O iii] λ5007 fluxes in these reference spectra agreed with previous measurements, where available. Our new measurement of F ([O iii] λ5007) = 3.67 10−13 ergs s−1 cm−2 for 3C120 was × larger than that of Peterson et al. (1998a), who measured F ([O iii] λ5007) = 3.02 × 10−13 ergs s−1 cm−2. Since our spectra have improved greatly in quality since then, we adopt our new [O iii] λ5007 flux. We did not find a published absolute [O iii] λ5007
flux measurement for Mrk 1501, so for that source we adopt the flux measured in our average spectrum of the photometric data as the absolute [O iii] flux. Using a
40 χ2 goodness-of-fit estimator method to minimize the flux differences between the spectra (van Groningen & Wanders 1992), we then scaled each individual spectrum to the reference spectrum. These procedures yield an absolute flux-calibrated data set for each object from which to measure the mean AGN luminosity. In some spectra, we were unable to obtain a good fit due to changes in spectrograph focus, so we manually scaled these spectra instead. Figure 3.1 shows the calibrated mean and root mean square residual (rms) spectra of our five objects based on the calibrated
MDM spectra.
3.2.2. Photometry
To supplement our spectra, we obtained V -band imaging observations using the 70-cm telescope at several different observatories. We used the 70-cm telescope at CrAO with the AP7p CCD, which has 512 512 pixels with a 15′ 15′ field of × × view when mounted at prime focus. We obtained data from the 46-cm Centurion telescope at Wise Observatory of Tel-Aviv University using an STL-6303E CCD with 3072 2048 pixels, with a field of view of 75′ 50′ for our setup. Further × × V -band observations were obtained using the University of Tokyo’s 1.0m miniTAO telescope stationed in Chile. We used the ANIR CCD camera, which has a pixel scale of 0.34′′pixel−1 and a field of view of 6 6 square arcseconds. Finally, we also × obtained observations using the SMARTS telescope in Chile.
41 3.3. Light Curves
3.3.1. Spectroscopic Light Curves
Emission-line light curves were created for both the MDM and CrAO data sets by fitting a linear continuum underneath the Hβ line in each spectrum and integrating the flux above it. The continuum was defined by two regions adjacent to the emission line, which is defined by regions given in Table 3.4. For the MDM data, the 5100 A˚ continuum light curves were created by taking the average flux measured in the wavelength regions listed in Table 3.4. Initial CrAO continuum and Hβ light curves were created the same way — however, the CrAO spectra were on a different
flux scale than the MDM spectra because different amounts of [O iii] and host galaxy light enter their slits due to changes in seeing, slit orientation, and aperture size. We assumed there is no real variability on timescales of less than 0.5 days, so we calibrated the CrAO light curves to the MDM light curves by multiplying the
fluxes by a constant calculated by taking the average flux ratios between pairs of observations from the CrAO and MDM light curves that are separated by less than
0.5 days, putting both light curves on the same flux scale.
42 3.3.2. Photometric Light Curves
For the WISE imaging data, we used image subtraction to produce the light curves using ISIS (Alard & Lupton 1998; Alard 2000). We generally follow the procedures of Shappee & Stanek (2011). The images are first aligned using a program called Sexterp (Siverd 2012, in prep). Sexterp is a replacement for ISIS’ default interp.csh that relies on SExtractor (Bertin & Arnouts 1996) for source identification. SExtractor source lists are significantly more robust and improve registration accuracy. We additionally use an upgraded interpolation utility provided with Sexterp. This routine implements the publicly available Bspline interpolation code of Th´evenaz et al. (2000) and produced better results with our images.
We then used ISIS to create a reference image for each field using the 20–30 images with the best seeing and lowest background counts. When creating the reference image, ISIS convolves the images with a spatially variable convolution kernel to transform all images to the same point-spread function (PSF) and background level. The resulting images are then stacked using a 3σ rejection limit from the median. We then used ISIS to convolve the reference image with a kernel to match it to each individual image in the data set and subtract each individual frame from its corresponding convolved reference image. We then extract light curves for the nucleus of each galaxy using ISIS to place a PSF-weighted aperture over the nucleus and measure the residual flux. We used varying extraction apertures
43 for the different objects, choosing apertures large enough to account for all AGN light but minimizing the host galaxy light included. For the CrAO images, we used photometric fluxes based on standard aperture photometry, which were measured within an aperture of 15′′.0. This includes all of the host galaxy flux for most of our objects, and was chosen to minimize slit losses due to variable seeing. See Sergeev et al. (2005) for more details on obtaining the CrAO photometric fluxes.
3.3.3. Combined Light Curves
The spectroscopic continuum light curves were merged with the photometric light curves as follows. We applied a multiplicative scale factor as well as an additive
flux adjustment to each photometric light curve to put them all on the same scale and correct for the differences in host galaxy starlight that enters the apertures (see
Peterson et al. 1995). The final continuum and emission-line light curves, scaled to our MDM light curves, are shown in Figures 3.2, 3.3, and 3.4. The continuum and Hβ fluxes are given in Appendix A and labeled according to the observatory at which they were obtained. Final light curve statistics are given in Tables 3.5 and 3.6.
3.4. Time Series Measurements
Previous reverberation studies have relied on fairly simple cross-correlation methods to measure the time delay between the continuum and emission-line
44 variations, τ. Recently, however, Zu et al. (2011) introduced an alternative method of measuring reverberation time lags called Stochastic Process Estimation for AGN
Reverberation (SPEAR1), and demonstrated its ability to recover accurate time lags.
We utilize this method here. As with cross correlation, we assume all emission-line light curves are scaled and shifted versions of the continuum light curve. SPEAR differs from simple cross correlation methods in two basic respects. First, SPEAR explicitly builds a model of the light curve and transfer function and fits it to both the continuum and the line data, maximizing the likelihood L of the model and then computing uncertainties using the (Bayesian) Markov Chain Monte Carlo method. Second, as part of this process it models the continuum light curve as an autoregressive process using a damped random walk (DRW) model. It has long been known that AGN continuum variability can be modeled as an autoregressive process (Gaskell & Peterson 1987) and a DRW model has been demonstrated to be a good statistical model of quasar variability using large ( 104) samples of quasar ∼ light curves (e.g., Kelly et al. 2009; Koz lowski et al. 2010; MacLeod et al. 2010; Zu et al. 2013). The parameters of the DRW model are included in the fits and their uncertainties, as is a simple top-hat model of the transfer function and the light curve means (or trends if desired).
The key physical advantage of SPEAR is that it automatically includes a self-consistent, physical model of how to interpolate in time. For any given DRW
1 Available at http://www.astronomy.ohio-state.edu/~yingzu/spear.html
45 model parameters, the stochastic process model gives a mathematical estimate for the light curve at any time along with its uncertainties that naturally includes all the information in both the continuum and line light curves and their uncertainties.
Since the DRW parameters also have to be estimated from the data, we allow them to vary as part of the overall model as well. In essence, this leads to a lag estimate that naturally includes the uncertainties in how to interpolate between data points, constrained by the physical properties of the variability in the target. Because it is then a statistical fit to the data with a set of parameters and a standard likelihood function, it also allows the use of powerful statistical methods like Markov Chain
Monte Carlo methods to produce uncertainties that correctly incorporate the effects of the model uncertainties on the lag estimate.
We used SPEAR on our light curves using the code described by Zu et al.
(2011) and successfully measured time lags (τSPEAR) for all five objects. We list these in Table 3.7. The mean and variance of the light curve models calculated by
SPEAR that are consistent with the data are shown in Figures 3.5, 3.6, and 3.7.
We also show the log-likelihood functions (log (L /Lmax) as a function of τ) for these light curves in Figure 3.8. The likelihood L is defined in Equation (17) in
2 Zu et al. (2011), and is proportional to e−χ /2. The best model, corresponding to
2 2 L , is associated with the minimum χ2, χ2 . Thus, L /L e(−(χ −χmin)/2) max min max ∝ and ∆χ2 = 2 ln(L /L ). Therefore, Figure 3.8 effectively shows ∆χ2 between − max models using each lag and the best model.
46 For comparison with previous results, we also include in Table 3.7 the lag measurements made using the interpolation method originally described by Gaskell
& Sparke (1986) and Gaskell & Peterson (1987) which was later modified by White &
Peterson (1994) and Peterson et al.(1998, 2004). We cross-correlate the continuum with the emission-line light curve, calculating the value of the cross correlation coefficient r at each each of many potential time lags. We show the CCFs for our light curves in Figure 3.8. Uncertainties in these lags are calculated using Monte
Carlo simulations that employ the flux randomization and random subset selection methods of Peterson et al. (1998a), as refined by Peterson et al. (2004). For each realization, we measure the lag (τpeak,CCF) that results in the peak value of the cross correlation coefficient, rpeak. We also measure the lag at the centroid of the CCF
(τcent,CCF), calculated using points surrounding the peak with values greater than
0.8rpeak. We adopt the mean of the distribution of delay measurements from our
Monte Carlo realizations, and the standard deviations of the same distributions are adopted as our formal 1σ uncertainties. In the cases of Mrk 335, Mrk 6, and
PG 2130+099, we subtracted linear trends before performing the CCF analysis, as there are clear secular trends in these light curves. This did not significantly affect the measured lag values, as can sometimes be the case. However, the resulting CCFs were cleaner, with much more narrow and well-defined peaks when the trends were subtracted.
47 3.4.1. Line Width and MBH Calculations
To determine the best value of the line width and its uncertainty, we use
Monte Carlo simulations similar to those used when determining the lag from the CCF. We run 100 simulations in which we create a mean and rms residual spectrum from a randomly chosen subset of the spectra, obtaining a distribution of resolution-corrected line widths. We take the mean value of σline or FWHM from these realizations and use their standard deviation as our uncertainty. We measure
σline and FWHM in both the mean and rms residual spectra for completeness, and report these in Table 3.8. We use the rms residual spectrum line widths to estimate
MBH, as this eliminates contamination from constant narrow line components and isolates the broad emission components that are actually responding to the continuum variations.
We adopt f = 5.5. This estimate is based on the assumption that AGNs h i follow the same MBH–σ∗ relationship as quiescent galaxies (Onken et al. 2004), and is consistent with Woo et al. (2010). This factor allows for easy comparison with previous results, but is about a factor of two larger than the value of f computed h i by Graham et al. (2011). We use σline(rms) in our MBH computation because there is at least some evidence that it produces less biased MBH measurements than using the FWHM (Peterson 2011). Using τSPEAR for the average time lag, we compute the
48 2 virial product (Mvir = cτ∆V /G) and MBH for all five galaxies. The measurements are reported in Table 3.8.
3.5. Discussion
3.5.1. The Radius–Luminosity Relationship
We compute the average 5100 A˚ luminosities of our sources, correcting for host galaxy contamination following Bentz et al. (2009a). We measure the observed-frame host-galaxy flux in our aperture for each source using HST images (Table 3.3).
With these measurements, we calculate the host-subtracted, rest-frame 5100 A˚
AGN luminosity for placement on the radius-luminosity relationship. The final host-subtracted AGN luminosities are given in Table 3.8. Note that we do not currently have HST images from which to measure the host luminosity for two of our objects, Mrk 6 and Mrk 1501. As a consequence, the luminosities listed for these objects are the total 5100 A˚ luminosities rather than just that of the AGN, and we expect them to fall to the right of the RBLR–L relationship.
Figure 3.9 shows the Bentz et al. (2009a) RBLR–L relationship and the placement of our new measurements. Previous measurements from Bentz et al.
(2009a) are represented as open shapes, while our new measurements are represented by filled shapes, varying in shape and color by object. We have not re-fit the best-fit
49 trend including our new data; we leave this to a future work. Mrk 335 and 3C120 both fall very close to their positions from the Bentz et al. (2009a), but we have increased the precision of their RBLR measurements. PG2130 +099 continues to lie somewhat to the right of the relation. Both Mrk 6 and Mrk 1501 also lie noticeably below the relationship, as is expected since we were unable to subtract the host galaxy starlight — we therefore show these luminosity measurements as upper limits.
Host measurements for these galaxies will shift both of them to lower luminosities and hence closer to the existing RBLR–L relation.
To see where we expect Mrk 1501 and Mrk 6 to lie on the relation after host subtraction, we examined the host galaxy light fraction in galaxies with similar BLR sizes (i.e. similar lags) as these two objects. Using measurements from Bentz et al.
(2009a), we calculated the average fraction of host galaxy light among galaxies with similar lags, and used this fraction to calculate the expected host galaxy fluxes, and hence the expected host-subtracted luminosities, in Mrk 1501 and Mrk 6. Host galaxies in objects with lags similar to Mrk 1501 contributed on average 34% of the total luminosity, so we expect Mrk 1501 to change from log λL = 44.32 0.05 5100 ± to around 44.10. Host galaxies in objects with lags similar to Mrk 6 contributed on average 56% of the total luminosity. If we applied this to Mrk 6, the host-subtracted luminosity would then be log λL5100 = 43.40. Both of these objects will likely continue to lie below the current RBLR–L relation, but within the normal range of scatter currently observed. However, it is important to note that there is a very large
50 scatter in the fraction of the luminosity contributed by the host galaxies in general, so these numbers are used for very rough estimations only.
3.5.2. Comments on Individual Objects
Mrk 335
Previous reverberation measurements of Mrk 335 were made by Kassebaum et al. (1997) and Peterson et al. (1998a) and reanalyzed by Peterson et al. (2004) and Zu et al. (2011). Previous Hβ measurements for this object are quite good, and it was included in this study mainly for the potential to measure the size of the high ionization component of the BLR. Details from our study have been reported by Grier et al. (2012a), and the data have been included in this study for
+0.4 completeness. Our new measurement of RBLR = 14.1−0.4 days is consistent with the
+3.6 previous measurement of RBLR = 15.3−2.2 (Zu et al. 2011) when taking into account the luminosity change of Mrk 335 between these two campaigns. In other words, the position of Mrk 335 on the RBLR–L relationship changed predictably given the expected photoionization slope of R L1/2 (i.e., τ L1/2). ∼ ∼
Mrk 1501
No previous reverberation mapping measurements exist for Mrk 1501. We
+2.2 measure τ = 15.5− days and a resulting black hole mass of M = (1.84 1.9 BH ±
51 8 0.27) 10 M⊙. As noted above, this object lies noticeably to the right of the ×
RBLR–L relation, which is expected since we have not yet subtracted the host galaxy contribution to the 5100 A˚ luminosity due to the lack of HST imaging data. As mentioned above, once we have corrected for host subtraction we expect the object to lie below the relation, but still within the normal scatter.
3C 120
3C 120 was observed by Peterson et al. (1998a) and reanalyzed by Peterson
+22.1 et al. (2004). The latter study reported τcent = 39.4−15.8 days, corresponding to
+3.14 7 M = 5.55− 10 M⊙. We included 3C 120 in our campaign in an effort BH 2.25 × +1.1 to reduce the large uncertainties in RBLR. Our new measurement of τ = 27.2−1.1
7 days leads to M = (6.7 0.6) 10 M⊙, which is consistent with the previous BH ± × measurements, but has much smaller uncertainties due to both better-sampled light curves and the improved techniques of measuring lags using SPEAR. Our new measurements place this object slightly below the RBLR–L relation, consistent with its previously-measured position.
Mrk 6
Mrk 6 was observed in reverberation studies by Sergeev et al. (1999),
Doroshenko & Sergeev (2003), and Doroshenko et al. (2012), who measured Hβ time lags using cross correlation. Doroshenko et al. (2012) report τ = 21.1 1.9 days. cent ± 52 8 This measurement was used to calculate M = (1.8 0.2) 10 M⊙. This study BH ± × used light curves that cover a very long time period with more sparse sampling than our campaign. Because of our dense time sampling, our light curves are sensitive to lags as small as a day or two. We measure a Hβ time lag of 9.2 0.8 days and M ± BH 8 = (1.36 0.13) 10 M⊙. ± ×
Our new τ measurement is substantially lower than the previous measurement
– however, varying BLR sizes are expected if the luminosity of the object changes, in accordance with the RBLR–L relation. In this case, the previous study reports lower
AGN luminosity measurements than we find, and by the RBLR–L relation we would also expect a smaller τ measurement in their data. However, they measure a lag on order of twice the length of ours, so this difference cannot be explained by a change in luminosity state. To investigate, we ran the light curves from Doroshenko et al.
(2012) through both the CCF and SPEAR analysis software, and obtain results that are generally consistent with theirs to within errors when using cross correlation.
However, we do note that the lags we measure using SPEAR are noticeably lower than the lags they report when we confine our attention to their more well-sampled light curves. For example, with their best-sampled light curves that cover the end
+1.2 of their observing period, we measure τ = 11.5−0.8 days, where they report τ =
4.6 20.4−4.1 days for the same light curves. The median spacing between observations in the Doroshenko et al. (2012) light curves is always above 10 days, which we suspect renders their light curves insensitive to lags shorter than this. We are confident that
53 our measurement of τ = 9.2 days is accurate for our data set, as the lag signal is clearly visible in our light curves and the sampling rate is very high in both the continuum and Hβ light curves.
Mrk 6 has a very interesting Hβ profile (see Figure 3.1) that has been observed to change dramatically both in flux and shape (Doroshenko & Sergeev 2003, Sergeev et al. 1999). The rms line profile from our study is clearly double-peaked and shows significant blending of the He ii emission with the Hβ emission. To verify that our line width measurement is not affected by the He ii component, we fit a second-order polynomial to the He ii feature in the rms spectrum and subtracted it from the total rms spectrum. We then re-measured the line width from this new spectrum and obtained a measurement consistent with that taken from the entire rms spectrum.
This suggests that the He ii blending did not affect our measurement of σline, so we adopted our original measurement for use in the MBH calculations. There are a variety of physical models that can produce this double-peaked profile, many of which we expect would show clear velocity-resolved signatures in our data. This analysis is beyond the scope of this paper and will be explored in detail in a future work.
PG 2130+099
Initial reverberation results for PG 2130+099 were first published by Kaspi et al. (2000), who measured a value of τ on the order of 200 days and thus
54 8 inferred a black hole mass of 1.4 10 M⊙. It was a significant outlier on both × the MBH–σ∗ and RBLR–L relations. However, PG 2130+099 was later re-observed
+4.4 7 and measured to have R = 22.9− days and M = (3.8 1.5) 10 M⊙ BLR 4.3 BH ± × (Grier et al. 2008), both of which are about an order of magnitude smaller than the original measurements. The discrepancy was attributed to undersampled light curves in the first measurements, as well as long-term secular changes in the Hβ equivalent width. While the 2008 data showed a clear reverberation signal, the amplitude of the variability in the study was quite low and the campaign was short in duration, rendering it insensitive to lags above 50 days, which made the light curves less than ideal. We included this object in our study in hopes of obtaining a better-sampled light curve sensitive to a wide range of time lags that would yield
+1.2 a more definitive result. Our new measurements of τ = 12.8−0.9 days and MBH
7 = (4.6 0.4) 10 M⊙ are consistent with those of Grier et al. (2008), but with ± × higher precision. Note that PG 2130+099 is in a noticeably different position on the
RBLR–L relation — it has moved nearly parallel to the relation from its previous location, since its luminosity has also changed. Like Mrk 335, this is consistent with the expectations from photoionization models of the BLR.
55 25 Mrk 335 6 Mrk 1501 Mean 20 Mean 4 15 10 2
2.5 rms 0.35 rms 2 0.3 0.25 1.5 0.2 1 0.15 4600 4800 5000 5200 5000 5200 5400 15 3C 120 20 Mrk 6 Mean Mean 10 15
10 5 0.8 5 0.7 rms 1.6 rms 0.6 1.4 0.5 0.4 1.2 0.3 4800 5000 5200 4600 4800 5000 5200
12 PG2130+099 10 Mean 8 6 4
rms 0.4
0.3
0.2 4600 4800 5000 5200 5400
Fig. 3.1.— The flux-calibrated mean and rms residual spectra of each object. All spectra are shown in the observed frame with the flux density in units of 10−15 erg s−1 cm−2 A˚−1. The dotted lines show the spectra before the [O iii] narrow emission lines have been subtracted, while the solid line shows the spectra after the subtraction. Note that we did not remove a narrow component of the Hβ emission line.
56 10
9
8
7
6 7 6.5 6 5.5 5
5450 5500 5550 2
1.8
1.6
1.4
2.6 2.4 2.2 2 1.8 1.6 5450 5500 5550
Fig. 3.2.— Complete light curves for Mrk 335 and Mrk 1501. For each object, the top panel shows the 5100 A˚ flux in units of 10−15 erg s−1 cm−2A˚−1 and the bottom panel shows the integrated Hβλ4861 flux in units of 10−13 erg s−1 cm−2. Open black circles denote the observations from MDM Observatory and red asterisks represent spectra taken at CrAO. Closed red squares show the photometric observations from CrAO, and closed blue triangles represent photometric observations from the WISE Observatory.
57 4.5
4
3.5
3 4.5
4
3.5
3 5450 5500 5550 12
10
8
8
7
6
5450 5500 5550
Fig. 3.3.— Complete light curves for 3C 120 and Mrk 6. See Figure 3.2 for details.
58 3.8 3.6 3.4 3.2 3 2.8 4.62.6
4.4
4.2
4
3.8
5450 5500 5550
Fig. 3.4.— Complete light curves for PG 2130+099. See Figure 3.2 for details.
59 10
9
8
7
6 7 6.5 6 5.5 5
5450 5500 5550 2
1.8
1.6
1.4
2.6 2.4 2.2 2 1.8 1.6 5450 5500 5550
Fig. 3.5.— The mean of the predicted light curves and their dispersions as estimated by the best-fit SPEAR model for Mrk 335 and Mrk 1501. For each object, the top panel shows the continuum light curve, and the bottom panel shows the Hβ light curve, both in the same units as Figures 3.2–3.4. The gray points show the merged light curves used in the model, and the solid line shows the mean of the SPEAR light curve models fit to the data. Dotted black lines show the standard deviation of values about the mean (see Zu et al. 2011).
60 4.5
4
3.5
3 4.5
4
3.5
3 5450 5500 5550 12
10
8
8
7
6
5450 5500 5550
Fig. 3.6.— The mean of the predicted light curves and their dispersions as estimated by the best-fit SPEAR model for for 3C 120 and Mrk 6. See Figure 3.5 for details.
61 3.8 3.6 3.4 3.2 3 2.8 4.62.6
4.4
4.2
4
3.8
5450 5500 5550
Fig. 3.7.— The mean of the predicted light curves and their dispersions as estimated by the best-fit SPEAR model for PG 2130+099. See Figure 3.5 for details.
62 1
0 0 0.8 0.9 -10
-10 0.6 0.8 -20
-20 0.4 0.7 -30
0.2 -30 -40 0.6
-50 0 -40 0.5 0 10 20 30 40 50 0 10 20 30 40 50
1 0 0
0.5
-10 -10 0.5
-20 0
-20
-30 0 -0.5 -30 -40
-50 -0.5 -40 -1 0 10 20 30 40 50 0 10 20 30 40 50
0.4
0
0.2
-20
0
-40 -0.2
-60 -0.4 0 10 20 30 40 50
Fig. 3.8.— Cross correlation and log-likelihood functions. The solid black lines show the log-likelihood functions from the SPEAR analyses, where the left axes show the SPEAR likelihood ratios log (L /Lmax). The dotted black lines show the cross correlation functions, whose r values are shown on the right axes. The ranges of the y-axes were chosen for easy comparison between the two curves.
63 Mrk 335 Mrk 1501 3C 120 Mrk 6 PG2130+099
Fig. 3.9.— The relationship between the BLR radius and AGN luminosity at 5100 A.˚ The calibration from Bentz et al. (2009a), is shown by the solid line. Gray squares are from Bentz et al. (2009a) and darker gray triangles are from Denney et al. (2010). Open colored shapes show previous measurements for our sources from Bentz et al. (2009a). The orange open square representing Mrk 6 is from Doroshenko et al. (2012). Filled colored shapes represent our new measurements of these objects. Each source was given its own shape and color combination for ease of comparison between the new and old measurements. Note that Mrk 6 and Mrk 1501 do not have their host galaxy starlight subtracted and therefore their continuum luminosities are shown as upper limits.
64 Object RA DEC z AB (J2000) (J2000) (mag)
Mrk335 000619.5 +201210 0.0258 0.153 Mrk1501 001031.0 +105830 0.0893 0.422 3C120 043311.1 +052116 0.0330 1.283 Mrk6 065212.2 +742537 0.0188 0.585 PG2130+099 21 32 27.8 +10 08 19 0.0630 0.192
Note. — Galactic extinctions are from Schlegel et al. (1998).
Table 3.1. 2010 Reverberation Campaign Object List.
65 Spectroscopy Photometry HJD HJD Object Observatory N ( 2450000) Observatory N ( 2450000) obs − obs −
Mrk335 MDM 78 5440-5559 CrAO 25 5431-5569 CrAO 7 5509-5568 WISE 19 5511-5545 Mrk1501 MDM 62 5440-5559 CrAO 63 5430-5568
66 CrAO 18 5443-5568 WISE 64 5433-5541 3C120 MDM 69 5441-5559 CrAO 64 5430-5568 CrAO 15 5456-5569 WISE 43 5436-5545 Mrk6 MDM 75 5441-5562 CrAO 59 5430-5569 CrAO 21 5443-5539 WISE 50 5435-5545 PG2130+099 MDM 68 5441-5557 CrAO 74 5430-5556 CrAO 20 5443-5539 WISE 72 5433-5541
Table 3.2. Spectroscopic and Photometric Observations FWHMa
Object [O iii] λ5007 F ([O iii]λ5007) FHost(5100 A)˚ (km s−1) (10−13 ergs s−1 cm−2) (10−15 erg s−1 cm−2 A˚−1)
Mrk335 280 2.31 0.10b 1.70 0.16 ± ± Mrk 1501 1.13 0.02c ··· ± ··· 3C 120 3.67 0.07c 0.685 0.063 ··· ± ± Mrk6 475 7.17 0.12c ± ··· PG2130+099 350 1.36 0.10d 0.601 0.055 ± ±
Note. — References: (a) Whittle (1992), (b) Peterson et al. (1998a), (c) This work, (d) Grier et al. (2008).
Table 3.3. Spectral Properties of Targets
67 Object Continuum Hβ Integration Integration Region (A)˚ Region (A)˚
Mrk335 5215-5240 4910-5100 Mrk1501 5540-5560 5190-5540 3C120 5250-5295 4930-5140 Mrk6 5140-5175 4820-5140 PG2130+099 5420-5435 5085-5284
Note. — All integration regions are in the observed frame.
Table 3.4. Continuum and Emission-Line Integration Regions
68 Sampling (days) Mean Objects T T Flux1 F R h i median var max (1) (2) (3) (4) (5) (6)
Mrk335 1.1 0.96 7.49 1.01 0.13 1.57 0.04 ± ± Mrk1501 0.66 0.48 1.49 0.16 0.11 1.48 0.04 ± ± 3C120 0.72 0.53 3.37 0.38 0.11 1.49 0.07 ± ± Mrk6 0.84 0.58 8.93 1.14 0.13 1.65 0.04 ± ± PG2130+099 0.55 0.43 3.1 0.23 0.07 1.33 0.03 ± ±
1Continuum and emission-line fluxes are given in 10−15 erg s−1 cm−2 A˚−1 and 10−13 erg s−1 cm−2, respectively, and have not been corrected for host galaxy contamination.
Note. — Column (1) lists the object, Columns (2) and (3) list the average and median time spacing between continuum observations, respectively. Column (4) gives the mean flux of the continuum in the observed frame. Column (5) gives the excess variance, defined by
√σ2 δ2 F = − (3.1) var f h i where σ2 is the flux variance of the observations, δ2 is the mean square uncertainty, and f is the mean observed flux (Rodriguez-Pascual et al. 1997). Column (6) is h i the ratio of the maximum to minimum flux in each light curve. Columns 7-11 are the same quantities but computed using the merged Hβ light curves rather than the continuum light curves.
Table 3.5. Continuum Light Curve Statistics
69 Sampling (days) Mean Objects T T Flux1 F R h i median var max (1) (2) (3) (4) (5) (6)
Mrk335 1.5 1.00 5.74 0.55 0.09 1.35 0.03 ± ± Mrk1501 1.55 0.99 2.16 0.23 0.10 1.44 0.05 ± ± 3C120 1.5 0.99 3.75 0.24 0.06 1.28 0.04 ± ± Mrk6 1.2 0.99 7.00 0.69 0.10 1.42 0.04 ± ± PG2130+099 1.32 0.99 4.17 0.17 0.04 1.14 0.02 ± ±
1Continuum and emission-line fluxes are given in 10−15 erg s−1 cm−2 A˚−1 and 10−13 erg s−1 cm−2, respectively, and have not been corrected for host galaxy contamination.
Note. — Columns are the same as Table 3.5.
Table 3.6. Hβ Light Curve Statistics
Object τSPEAR τcent,CCF τpeak,CCF (days) (days) (days) (1) (2) (3 (4)
+0.4 Mrk335 14.1−0.4 14.3 0.7 14.0 0.9 +2.2 ± ± Mrk1501 15.5−1.8 12.6 3.9 13.8 5.4 +1.1 ± ± 3C120 27.2−1.1 25.9 2.3 25.6 2.4 +0.8 ± ± Mrk6 9.2−0.8 10.1 1.1 10.2 1.2 +1.2 ± ± PG2130+099 12.8− 9.6 1.2 9.7 1.3 0.9 ± ±
Table 3.7. Rest-frame Hβ Lag Measurements
70 Object σline(mean) FWHM(mean) σline(rms) FWHM(rms) Mvir MBH log λL5100 −1 −1 −1 −1 6 6 −1 (km s ) (km s ) (km s ) (km s ) ( 10 M⊙) ( 10 M⊙) (erg s ) × × (1) (2) (3) (4) (5) (6) (7) (8)
Mrk335 1663 6 1273 3 1293 64 1025 35 4.6 0.5 25 3 43.70 0.08
71 ± ± ± ± ± ± ± Mrk1501 3106 15 3494 35 3321 107 5054 145 33.4 4.9 184 27 44.32 0.05 ± ± ± ± ± ± ± 3C120 1687 4 1430 16 1514 65 2539 466 12.2 1.2 67 6 43.96 0.06 ± ± ± ± ± ± ± Mrk6 4006 6 2619 24 3714 68 9744 370 24.8 2.3 136 12 43.75 0.06 ± ± ± ± ± ± ± PG2130+099 1760 2 1781 5 1825 65 2097 102 8.3 0.7 46 4 44.15 0.03 ± ± ± ± ± ± ±
Table 3.8. Mean and RMS Line Widths, Virial Masses and Luminosities. Chapter 4
The High-Ionization Component of the BLR in Mrk 335
4.1. Background
Narrow-line Seyfert 1 galaxies (NLS1s) are a subset of AGNs that show narrower broad emission-line components than typical Type 1 AGNs, as well as a number of other distinguishing properties (Osterbrock & Pogge 1985, Goodrich
1989, Boller et al. 1996). Explanations for their unique characteristics include the possibility that they are either low-inclination or high-Eddington rate accreters (or both – see Boroson 2011). Substantial blue enhancements in high ionization lines such as C ivλ1549 and He iiλ1640 are apparently typical of, although not restricted to, NLS1 galaxies (e.g., Richards et al. 2002, Sulentic et al. 2000). This may be evidence for material in a disk wind (Richards et al. 2011, Leighly & Moore 2004).
If the blue enhancement is due to a wind, the use of high ionization emission lines to measure virial black hole masses (MBH) in these objects may be problematic, as the method relies on the assumption that the emitting gas is in virial motion around the black hole.
72 If broad high-ionization lines like He ii λ4686 are emitted from virialized gas near the black hole, we expect much shorter reverberation time lags for He ii λ4686 than for low ionization lines like Hβλ4861 because both ionization stratification and the line width require this gas to be much closer to the central source. Peterson et al.
(2000) investigated the NLS1 NGC 4051 and detected very broad, blue-enhanced
He ii λ4686 emission in the RMS spectrum. Unfortunately, their time resolution was inadequate to reliably measure a He ii lag. Denney et al. (2009a) measured an improved Hβ lag for NGC 4051 using data from their 2007 campaign, but were similarly unable to recover a He ii lag. Bentz et al. (2010a) report marginal detections of He ii lags in two NLS1 galaxies observed in their 2008 observing campaign. These recent studies achieved similar high sampling rates, but the expected He ii lags for these targets are too short for a robust detection in these datasets. We have therefore been unable to test whether this emission originates in outflowing gas or is in virial motion.
We included the NLS1 galaxy Mrk 335 our recent reverberation mapping campaign, which is described in detail in Chapter 3. One goal of this high sampling rate program was to measure the reverberation lag for a high-ionization line,
He ii λ4686, in this source. In this chapter, we present the He ii results for Mrk 335, having obtained a high enough sampling rate to measure its short time delay, and show that the MBH estimate from the high ionization He ii line agrees with those from low ionization lines.
73 4.2. Observations
Details on observations and subsequent analysis techniques were discussed in Chapter 3 and published by Grier et al. (2012b). In general, we follow the observational and analysis practices of Denney et al. (2010), which largely follows the analysis described by Peterson et al. (2004). We briefly describe the observations and analysis techniques again below for completeness.
4.2.1. Spectroscopy
The majority of the spectra were obtained using the 1.3m McGraw-Hill telescope at MDM Observatory. We used the Boller and Chivens CCD spectrograph to obtain 82 spectra over the course of 120 nights from 2010 Aug 31 to Dec 28. We used the 350 mm−1 grating to obtain a dispersion of 1.33 A˚ pixel−1, with a central wavelength of 5150 A˚ and overall spectral coverage from roughly 4400 A˚ to 5850 A.˚
The slit was oriented north-south and set to a width of 5′′.0 and we used an extraction window of 12′′.0, which resulted in a spectral resolution of 7.9 A.˚ Figure 4.1 shows the mean and root mean square (RMS) spectra of Mrk 335 from the MDM spectra.
We also obtained 7 spectra with the Nasmith spectrograph and SPEC-10 CCD at the 2.6m Shajn telescope at the Crimean Astrophysical Observatory (CrAO). We used a 3′′.0 slit at a position angle of 90◦. Spectral coverage was from approximately
3900 A˚ to 6100 A.˚
74 4.2.2. Photometry
We collected 25 epochs of V -band photometry from the 70-cm telescope at
CrAO using the AP7p CCD at prime focus, covering a 15′ 15′ field of view. The × flux was measured within an aperture of 15′′.0. See Sergeev et al. (2005) for more details.
We also obtained 19 epochs of V -band photometry at the Wise Observatory of Tel-Aviv University using the Centurion 18-inch telescope with a 3072 2048 × STL-6303E CCD and a field of view of 75′ 50′. For these data, we used the ISIS × image subtraction package rather than aperture photometry to measure fluxes (Alard
& Lupton 1998; Alard 2000) following the procedure of Shappee & Stanek (2011).
4.3. Light Curves and Time Series Analysis
4.3.1. Light Curves
The reduced spectra were flux-calibrated assuming that the [O iii] λ5007 emission line flux is constant (see Denney et al. 2010 for details on data processing).
Emission-line light curves were created for both the MDM and CrAO data sets by
fitting linear continua underneath the Hβ and He ii lines and integrating the flux above them. Hβ fluxes were measured between 4910–5100 A˚ in the observed frame, with the continuum interpolation defined by the regions 4895–4910 and 5215–5240
75 A.˚ He ii λ4686 fluxes were measured from 4660–4895 A,˚ with the continuum defined from 4550–4575 and 4895–4910 A.˚ The CrAO light curves were then scaled to the
MDM light curve to account for the different amounts of [O iii] light that enters the slits due to differences in seeing, slit orientation, and aperture size.
The continuum light curve was created by taking the average 5100 A˚ continuum
flux of the MDM spectra, measured between 5215-5240 A˚ in the observed frame. This light curve was then scaled and merged with the other continuum and photometric light curves with corrections for the host galaxy starlight in the different apertures
(see Peterson et al. 1995). The final continuum and emission-line light curves are shown in Figure 4.2. Light curve statistics are given in Table 4.1.
4.3.2. Time delay measurements
For comparison with previous results, we used the interpolation method originally described by Gaskell & Sparke (1986) and Gaskell & Peterson (1987) which was later modified by White & Peterson (1994) and Peterson et al.(1998,
2004) to measure the time lag. We cross-correlated the two light curves with one another, calculating the value of the cross correlation coefficient r at each value of time lag. Figure 4.3 shows the resulting cross correlation functions (CCFs) for the light curves. Uncertainties in lags are calculated using Monte Carlo simulations that employ the methods of Peterson et al. (1998a) and refined by Peterson et al.
76 (2004). For each realization, we measure the location of the peak value of the cross correlation coefficient (τpeak,CCF), and the centroid of the CCF (τcent,CCF), calculated using points surrounding the peak. We adopt the mean τpeak,CCF and τcent,CCF from the Monte Carlo realizations for our delay measurements and the standard deviation as our formal uncertainties. Before the light curves were cross correlated, we removed the long-term linear upward trend that is clearly visible in all three light curves (see
Figure 4.2). Welsh (1999) discusses the value in this practice of “detrending” the light curves, as the cross correlation function (CCF) tends to latch onto long-term trends unassociated with reverberation, often resulting in incorrect lags. From our cross correlation analysis, we measure τ (Hβ) = 13.9 0.9 days and cent,CCF ± τ (He ii) = 2.7 0.6 days. All lag measurements are listed in Table 4.2. cent,CCF ±
Previous reverberation studies have relied on these fairly simple cross correlation methods to measure τ. Recently, however, Zu et al. (2011) discussed an alternative method of measuring reverberation time lags, originally called the Stochastic
Process Estimation for AGN Reverberation (SPEAR) and demonstrated its ability to recover accurate time lags. We discussed this method in more detail in Chapter
3. The basic idea is to assume all emission-line light curves are scaled and shifted versions of the continuum light curve. One then fits the light curves using a damped random walk model (e.g. Kelly et al. 2009, Koz lowski et al. 2010, MacLeod et al.
2010) and then aligns them to determine the time lag. Uncertainties in lags are computed using a Markov Chain Monte Carlo method (see Zu et al. 2011). SPEAR
77 is remarkably good at predicting time lags in data sets with relatively large gaps in the sampling. Using SPEAR, we successfully recover time lags for both the Hβ and
He ii emission lines. We allowed SPEAR to automatically remove the linear trend and include any resulting uncertainties in the overall lag uncertainties. We measure
+0.7 τ (Hβ) = 14.0 0.3 days and τ (He ii)=1.6− days, also reported in SPEAR ± SPEAR 0.5 Table 4.2. We see good agreement with the CCF results from the Hβ emission line, and while there is a small difference between SPEAR and CCF lags for the He ii line, they are still statistically consistent with one another. We suspect this small difference is due to the gap in data that is very close to the peak in the He ii light curve. We adopt the CCF values for our mass calcuations to allow comparison with previous reverberation efforts.
4.3.3. Line width measurement and MBH calculations
Assuming that the motion of the Hβ-emitting gas is dominated by gravity, the
2 relation between MBH, line width, and time delay is MBH = fcτ∆V /G, where τ is the measured emission-line time delay, ∆V is the velocity dispersion of the BLR, and f is a dimensionless factor that accounts for the structure within the BLR. The
BLR velocity dispersion can be estimated using the line width of the measured broad emission line in question. This width is usually characterized by either the FWHM or the line dispersion, σline. We use σline because there is evidence that it produces less biased MBH measurements (Peterson 2011). We measure the line width in the
78 RMS spectrum, which eliminates contributions from the contaminating narrow components. We adopt an average value of
To determine the best value of σline, we use Monte Carlo simulations following
Peterson et al. (2004). The resulting line widths are given in Table 4.2. Using our measured values of τcent,CCF for the average time lag and σline from the RMS spectrum as ∆V , we compute MBH using both the Hβ and He ii emission lines.
7 We measure M = (2.7 0.3) 10 M⊙ using the Hβ emission line and M = BH ± × BH 7 (2.6 0.6) 10 M⊙ using He ii. ± ×
4.4. Discussion
As discussed above, several studies involving NLS1 galaxies have found indications of outflows in high-ionization lines in the form of enhanced flux on the blue side of the emission lines. Inspection of Mrk 335 spectra from the HST archive shows this enhanced blueward flux is present in the C ivλ1549 line as well, but the
He iiλ1640 line is blended with C iv, so we cannot see if it too exhibits this blue enhancement. In fact, the shape of the He ii λ4686 emission line in the RMS spectrum
79 of Mrk 335 (Figure 4.1) shows red and blue shoulders that could be a signature of disk structure. To search for possible outflow signatures in the He ii λ4686 emission line, we divided it into red and blue components and integrated each component separately, creating two He ii light curves. We then cross-correlated the red and blue light curves with one another to see if there is any time delay between the two components. Cross correlation analysis yields a centroid lag τ = 0.4 0.8 days. cent ± This is consistent with zero and thus presents no evidence for bulk outflows in the
He ii λ4686 emission of Mrk 335. The consistency of the MBH measurements made using the He ii lines with those from Hβ are also suggestive of virial motion rather than outflowing gas.
Previous reverberation measurements of Mrk 335 were made by Kassebaum et al. (1997) and Peterson et al. (1998a) and subsequently reanalyzed by Peterson
+3.6 et al. (2004) and Zu et al. (2011). Zu et al. (2011) report a time delay of 15.3−2.2 days and σ 920 km s−1 for Hβ, but were unable to make a robust He ii line ∼ measurement, as their average time sampling was on the order of 10 days. Peterson
7 et al. (2004) measure M = (1.4 0.4) 10 M⊙ from the Hβ emission line. Our BH ± ×
MBH measurements deviate from theirs by almost a factor of two. We suspect the difference in MBH is due to the difference in the line width measurements between the two campaigns. The uncertainties in line width measurements and in the f factor are the main sources of uncertainties in reverberation MBH measurements – when the light curves are well-sampled, the lag measurements themselves have been
80 shown to be remarkably robust (e.g. Watson et al. 2011). Peterson et al. (2004)
(their Table 6) find that the virial products computed for an object using data from different epochs often differ from one another by as much as a factor of two (e.g.
NGC 5548) and that the typical fractional error in the virial products is about 33%.
Given our uncertainties in the f factor and the limitations in trying to accurately describe the BLR velocity field with a single line-width characterization, we probably cannot actually do better than about a factor of two or three in individual MBH measurements.
Ground-based reverberation campaigns in the past have been limited to objects with Hβ time lags that are expected to be less than a month or two due to both the finite length of the campaigns (which typically last 50–100 days) and the fact that most objects are only observable from the ground for only about half of the year. Measurement of longer time lags would require extended campaigns, which are difficult to schedule. If, as our evidence suggests, the He ii emission line is in virial motion around the black hole, we can use this emission line to measure MBH in objects at higher redshifts, as expected He ii lags in many of these high-luminosity objects are short enough to measure in one observing season.
81 25
20
15
10
Fig. 4.1.— The flux-calibrated mean (top) and RMS (bottom) spectra of Mrk335 in the observed frame (z = 0.02579). The flux density is in units of 10−15 erg s−1 cm−2 A˚−1. The dashed lines show the spectra before the [O iii] narrow emission lines were removed; the solid line shows the spectra after the subtraction.
82 10
9
8
7
6
4
3.5
3
2.5
2
HJD (-2450000)
Fig. 4.2.— Complete light curves for Mrk 335 from our observing campaign. The top panel shows the 5100A˚ flux in units of 10−15 erg s−1 cm−2A˚−1, the middle panel shows the flux in the He ii λ4686 region in units of 10−13 erg s−1 cm−2, and the bottom panel shows the integrated Hβλ4861 flux, also in units of 10−13 erg s−1 cm−2. Open circles denote observations from MDM Observatory and asterisks represent spectra taken at CrAO. Closed squares show the photometric observations from CrAO, and closed triangles represent photometric observations from WISE Observatory. Vertical dashed lines have been placed at two obvious features in the continuum to aid the eye. The vertical lines have been shifted by the measured He ii and Hβ lag values (2.7 days and 13.9 days, respectively) to aid the eye in identifying the correct lag values for each emission line. Dotted lines show the trends that were subtracted before performing the cross correlation analysis.
83 Fig. 4.3.— CCFs for the light curves. Dashed lines represent the autocorrelation function (ACF) of the continuum light curve, and solid lines show the CCFs for the emission lines.
84 Sampling Mean Time Interval (days) Mean Fractional Series N T T Flux Error F R h i median var max (1) (2) (3) (4) (5) (6) (7) (8)
5100 A˚ 133 1.0 0.95 7.57 1.01 0.013 0.129 1.53 0.04 ± ± He ii λ4686 89 1.5 1.00 7.65 1.00 0.016 0.130 1.55 0.04 ± ± Hβλ4861 89 1.5 1.00 5.74 0.53 0.015 0.091 1.38 0.03 ± ±
Note. — Column (1) lists the spectral feature, and column (2) gives the number of points in the individual light curves. Columns (3) and (4) list the average and median time spacing between observations, respectively. Column (5) gives the mean flux of the feature in the observed frame , and column (6) shows the mean fractional error that is computed based on observations that are closely spaced in time. Column (7) gives the excess variance, defined by
√σ2 δ2 F = − (4.1) var f h i where σ2 is the flux variance of the observations, δ2 is the mean square uncertainty, and f is the mean observed flux. Column (8) is the ratio of the maximum to h i minimum flux in each light curve. ∗Continuum and emission-line fluxes are given in 10−15 erg s−1 cm−2A˚−1 and 10−13 erg s−1 cm−2, respectively.
Table 4.1. Mrk 335 Light Curve Statistics
85 Parameter Hβ He ii (1) (2) (3)
τ a 13.9 0.9 days 2.7 0.6 days cent,CCF ± ± b τpeak,CCF 13.8 0.8 days 2.1 1.2 days ± ±+0.7 τ 14.0 0.3 days 1.6 − days SPEAR ± 0.5 σ (mean) 1641 12 km s−1 3465 26 km s−1 line ± ± FWHM (mean) 1363 15 km s−1 3191 571 km s−1 ± ± σ (RMS) 1336 51 km s−1 3001 277 km s−1 line ± ± FWHM (RMS) 1149 38 km s−1 7380 1275 km s−1 ± ± 7 7 M (2.7 0.3) 10 M⊙ (2.6 0.8) 10 M⊙ BH ± × ± ×
Note. — All values are given in the rest frame of the object.
Table 4.2. Hβ and He ii λ4686 Time Series Results
86 Chapter 5
The Structure of the BLR: Reconstructed Velocity-Delay Maps
5.1. Background
The ultimate goal of our 2010 reverberation campaign, described in detail in Chapter 3, was to obtain velocity-resolved reverberation results — namely, to recover velocity-delay maps, for several of the targets. In this chapter, we present velocity-binned reverberation results for all five objects observed in our 2010 campaign and two-dimensional velocity-delay maps for four objects. While our reverberation campaign was aimed primarily at investigating the Hβ emission line, in a few cases we recover velocity-delay maps for the Hγ and He ii λ4686 emission lines as well. In this chapter, the data discussed are the spectra and light curves reduced and processed as described in Chapter 3.
87 5.2. Initial Hβ Velocity-Binned Time Series Analysis
As a preliminary step, we first searched for gross kinematic signatures by seeing if different parts of each emission line show different time delays with respect to the continuum. Following Denney et al. (2009b), we divided the Hβ emission line into velocity bins as follows: First, we divide the Hβ emission lines in half at the zero-velocity line center (the systemic redshift), to separate red-shifted and blue-shifted signals. We then divide each line half into bins containing equal flux in the rms residual spectrum, choosing the number of bins for each object based on the width of the rms line profiles. We create light curves for individual bins by integrating the flux within each bin in each scaled spectrum. The resulting light curves were analyzed using JAVELIN, the updated version of SPEAR (see Zu et al.
2011 and Grier et al. 2012b for details), to measure the time delay in each light curve with respect to the continuum light curve. In essence, JAVELIN uses a statistical model of the continuum light curve and its covariances and simultaneously fits the continuum and line light curves assuming a simple top-hat transfer function. This leads to a statistically well-defined means of interpolating the irregularly sampled line and continuum data that essentially averages over all possible interpolated light curves weighted by their likelihood of fitting the data. In particular, the model fills gaps in the light curves in a well-defined manner with well-defined uncertainties.
88 Our results from JAVELIN are shown in Figure 5.1. The top panels show the rms residual line profile and the wavelength bins, and the bottom panel shows the distribution of lags for each part of the emission line. As a check, we also ran the light curves for each bin through a cross correlation analysis routine that is widely used in reverberation mapping (e.g., Peterson et al. 2004). The lags from the cross correlation were nearly identical to the ones obtained from JAVELIN. We show here only the JAVELIN results, since the JAVELIN uncertainties are much smaller than those obtained through cross correlation methods and the overall signatures are the same. In all five objects, we see velocity-dependent time lags in the Hβ emission line. Mrk 335, Mrk 1501, and PG 2130+099 all show longer time lags on the blue side of the line profile than on the red side, which is a signature of inflowing gas.
By contrast, 3C 120 shows a nearly symmetric profile, with small lags in the outer wings, and larger lags towards the line center, which is suggestive of a disk. Mrk 6 shows more complex structures that could be a combination of a disk plus infall, as previously suggested by Doroshenko et al. (2012) for this object.
5.3. Two-Dimensional MEMECHO Fitting
5.3.1. Preparation of Spectra
To recover the velocity-delay maps, we used maximum entropy methods as implemented in the program MEMECHO (see Horne et al. 1991 and Horne 1994 for
89 details). To prepare the data for use with MEMECHO, we used software developed by Keith Horne called PREPSPEC. PREPSPEC applies corrections to account for wavelength shifts due to instrument flexure and differential refraction, spectral blurring due to seeing and instrumental resolution, and errors due to slit losses and changes in atmospheric transmission between epochs. To do this, we fit a calibration model to the spectra that accounts for both the spectral variability and the abovementioned systematic errors. We model the spectra as the sum of five different components: 1) a constant “mean” spectrum, which accounts for non-varying components of the continuum and broad emission lines as well as starlight from the host galaxy, 2) non-varying narrow emission lines, 3) a time-variable continuum,
4) time variable broad emission lines, and 5) factors to account for wavelength shift, spectral blurring, and scaling. After subtracting the models from the spectra, the resulting continuum-subtracted line profile variations are used as inputs to
MEMECHO.
5.3.2. MEM Fitting
In brief, MEMECHO finds the “simplest” linearized echo model that fits the observed continuum and emission-line spectral variations. This is accomplished by minimizing the function Q = χ2 2 αS for a given model fit to a set of data. − 2 Here χ measures the “badness-of-fit” between N data points with values Dk and the corresponding model predictions µk for those values, assuming Gaussian errors
90 2 with known variances σk. The entropy S measures the “simplicity” of the model, elaborated below, and the regularization parameter α controls the trade-off between these competing requirements. MEMECHO adjusts α and the model parameters pi to achieve a user-specified value of χ2/N while maximizing S.
In MEMECHO’s linearized echo model
∞ L(λ, t)= L0(λ)+ Ψ(λ, τ)(C(t τ) C0)dτ, (5.1) Z0 − −
the parameters pi include the continuum light curve C(t) on an evenly spaced grid, the delay map at each wavelength Ψ(λ, τ), and the time-independent background spectrum L0(λ). C0 is a reference continuum level, which we set at the median of the continuum light curve data. The entropy of the model is defined as
S = wi(pi qi pi ln (pi/qi)), (5.2) Xi − −
where wi are weights, pi are the positive parameters outlined above, and qi are the default values of these parameters. Note that S is maximized when pi = qi.
Minimizing Q gives
N Dk µk dµk 0= −2 + αwi ln (pi/qi). (5.3) kX=1 σk dpi
Thus the model parameters pi are pulled by the data toward µk = Dk and by the entropy toward pi = qi, with α adjusting the trade-off between the two. With default
91 1/2 values qi = (pi−1 pi+1) , the entropy “pulls” each pi toward the geometric mean of its neighbors, so that the entropy penalizes regions of high curvature and favors smooth functions with gaussian features and exponential tails. The weights wi and default values qi are assigned with two parameters in MEMECHO: W to stiffen
Ψ(λ, τ) relative to C(t), and A to control the aspect ratio of features in Ψ(λ, τ).
The A and W parameters are set by the user. Their qualitative effects on the velocity-delay maps are described below in Section 5.4.2.
5.3.3. The Driving Continuum Light Curve
One of the major practical issues with the use of MEMECHO is its tendency to introduce spurious features at gaps in the continuum light curve to drive χ2/N to the minimum value set by the user. This becomes a significant problem as the target χ2/N decreases. Altering MEMECHO parameters to stiffen (i.e., penalize rapid variations in) the driving light curve is sometimes helpful, but we were still unable to produce velocity-delay maps with any discernible structure using the original continuum light curves. To provide stronger constraints on the driving light curve model, we ran our full continuum light curves (including both spectroscopic and photometric data) through the JAVELIN modeling software of Zu et al. (2011).
This method models the continuum with a damped random walk (DRW) model that has been demonstrated to be a good statistical model of AGN variability (e.g.,
Kelly et al. 2009; Koz lowski et al. 2010; MacLeod et al. 2010; MacLeod et al. 2012;
92 Zu et al. 2013). This allowed us to create highly sampled simulated continuum light curves with reliable uncertainties that represent the range of the most likely continuum behavior within the gaps. Instead of providing MEMECHO with our original continuum light curve, we use our highly sampled JAVELIN mean light curve, which more strongly constrains the MEMECHO continuum model. This light curve is the probability-weighted mean of DRW light curves consistent with the data and the DRW model. Uncertainties on each point in the simulated light curve represent the standard deviation of probable light curves around this mean. These narrow to match the measurement errors of the data points and then broaden as the gaps between the data become larger. The variability observed in the adopted mean simulated light curve is somewhat smoothed compared to an individual DRW model realization, but the error envelope applied to the mean light curve, representing the 1σ deviations of the individual realizations about the mean, accounts for these differences. The original continuum light curves and their corresponding simulated light curves are shown in Figure 5.2.
Using the JAVELIN mean continuum light curve allows us to fit the variability of both the continuum and the rest of the spectrum to much higher degrees of accuracy and keep MEMECHO from introducing spurious features into our light curves and by extension, into the derived echo maps. Light curves, 1-dimensional
MEMECHO fits, and recovered delay maps at selected wavelengths are shown in
Figures 5.3-5.7 for all five objects. The wavelengths for the fits and delay maps
93 shown in these figures were chosen to show the response in the red and blue wings as well as at the center of each emission line. One thing to note is that using the probability-weighted mean as the constraint on the MEMECHO construction of the continuum means that we are over-smoothing the light curves because we are neglecting the covariance of light curve deviations from the mean, which may weaken short-timescale variability. This will also cause some difficulty in the fitting, as it can force the light curve models to be smoother than the data. At the same time, the
χ2/N value reported by MEMECHO is no longer strictly valid — the deviations of any particular light curve model from the “mean” continuum light curve are highly correlated, so, for example, having 10 consecutive points 1σ from the light curve model is likely a 1σ deviation, not a 10σ deviation. However, the velocity-delay maps do not change significantly if we use random individual DRW realizations of the light curve instead of the mean, indicating that the mean light curve can be used to produce accurate velocity-delay maps. While this solution is not ideal, it is currently the best method we have of dealing with gaps in the observed continuum light curves pending a major effort to integrate the two distinct software packages
(JAVELIN and MEMECHO).
94 5.4. Two-Dimensional Velocity-Delay Maps
We were able to recover velocity-delay maps with MEMECHO for four out of the five galaxies observed. The best maps are for 3C 120 and Mrk 335, while the ones obtained for PG 2130+099 and Mrk 1501 are somewhat less well-defined. The light curves for Mrk 6, on the other hand, do not seem to be well-fit by a simple echo model, and we were therefore unable to obtain two-dimensional velocity-delay maps for this object. Figure 5.6 shows the best model fits we were able to obtain for Mrk 6. While there is some evidence for velocity-dependent structure in the one-dimensional delay maps shown, the models do not successfully fit any of the short-term variations seen in this object. We were unable to improve the light curve fits by lowering the target χ2/N, as MEMECHO was unable to converge on a solution when we did so.
The best-fit velocity-delay maps for the other four objects, covering the full observed wavelength range for each, are shown in Figure 5.8. The parameters used in MEMECHO to create the velocity-delay maps are given in Table 5.1. We also show more detailed velocity-delay maps for each individual emission line recovered for each object in Figures 5.9-5.12. To aid the eye in comparing ionization-stratified
BLR structure between different emission lines in each object, we also provide false-color maps in Figure 5.13, with each color representing a different emission line as described in the captions. We also created simulated velocity-delay maps for
95 a few different BLR kinematic models to compare qualitatively with our recovered velocity-delay maps. These simulated maps are shown in Figure 5.14, and represent
7 different BLR models around a black hole with M = 1 10 M⊙. We see features BH × reminiscent of these simple models in our recovered velocity-delay maps.
5.4.1. Comments on Individual Objects
Mrk 335
The velocity-binned analysis of Mrk 335 shows a definite velocity-dependent lag signature (Figure 5.1), and we clearly see similar structures in the velocity-delay maps of both the He ii and Hβ emission lines (Figure 5.9). We see a chevron-shaped pattern in the He ii line, with a lack of prompt response in the center and shorter delays in the wings, which is consistent with the signatures expected from an inclined disk or a spherical shell (Figure 5.14). In the Hβ emission, we see an asymmetric profile, with longer lags towards the blue end of the emission, and shorter lags towards the red end. This is suggestive of inflowing gas, as demonstrated in our simple infall models given in Figure 5.14. This also matches the signature we found in our initial velocity-binned analysis. The He ii emission is confined to smaller delays, as is expected from photoionization models of the BLR. Also consistent with disk structure is the response of He ii along a much wider velocity range than that of Hβ.
96 Mrk 1501
Fewer kinematic details are apparent in the recovered maps for Mrk 1501 than for the other objects in our sample. This may be a consequence of the noisier continuum and emission-line light curves obtained for this target. Nonetheless, we can still gain useful insights into the Hβ and Hγ-emitting regions of the BLR in this object. As with Mrk 335, we see evidence in the velocity-delay maps for inflow, with longer lags in the blue and shorter lags towards the red end of the Hβ and Hγ emission lines. This signature is closest to that of the “extended BLR” infall model shown in Figure 5.14. While this velocity-delay map is probably too low-resolution for any detailed modeling, we do see this same signature in our velocity-resolved analysis in Figure 5.1. There is also evidence for radial stratification of the BLR —
Figure 5.13 shows there is a much stronger Hβ response at longer time lags than is observed for the Hγ emission.
3C 120
Our velocity-delay maps for 3C 120, shown in full in Figure 5.8, are the cleanest maps we were able to recover. We see evidence for radial stratification, with He ii showing the shortest delays and Hγ and Hβ emitted at progressively larger radii.
Figure 5.13 shows all three emission lines on one scale, with Hβ, Hγ, and He ii shown in red, blue, and green, respectively, highlighting the radial stratification.
97 The lack of a prompt response at line center in both Balmer lines indicates a deficit of material along the line of sight. The shape of the hydrogen response is consistent with signatures expected from an inclined disk or a spherical shell (Figure 5.14).
This is similar to the signal found in NGC5548 (Horne et al. 1991), Arp 151 (Bentz et al. 2010b), and Mrk 50 (Pancoast et al. 2012). We also see an asymmetry in the strength of the response in both the He ii and Hβ emission profiles, nominally indicating inflow. The smaller velocity range of the Hβ and Hγ emission with respect to He ii, combined with the radial stratification signatures, is consistent with disk structure.
PG 2130+099
The velocity-delay map for PG 2130+099 is somewhat noisier than that of
3C 120, and the model fits are not as good. However, the structure of the map is worth noting. While reverberation delays for the He ii emission remain unresolved for this data, both the Hβ and Hγ emission lines clearly show velocity-resolved delay structure, with strong asymmetries as a function of velocity across the emission-line profile. This is similar to the Hβ emission we see in Mrk 335 and the He ii emission in 3C 120. This asymmetry, with longer lags at the blue end of the emission line and shorter lags to the red, is suggestive of inflowing gas, and matches the
“Infall (less-extended BLR)” model in Figure 5.14 quite well. In our analysis of the velocity-binned sections of the Hβ emission line (Figure 5.1), we see this same
98 structure, so while the delay maps for this object are probably not good enough for any detailed modeling, they are consistent with the signatures of inflowing material already seen in this object.
PG 2130+099 has long been a curiosity, as discussed in Chapter 2. Early on,
Kaspi et al. (2000) measured a time lag of 180 days in this object, placing it well ∼ above the RBLR–L relationship. Our later studies (Grier et al. 2008; Grier et al.
2012b) found much shorter lags on the order of tens of days: Most recently, we
+1.2 reported a mean Hβ time delay of 12.8−0.9 days. We attributed the discrepancies to undersampled light curves combined with long-term secular changes in the Hβ equivalent width in the data from Kaspi et al. (2000). However, even with the new, shorter lag measurements, PG 2130+099 is still a major outlier from the RBLR–L relation, as it is now positioned far below the relation (Grier et al. 2012b). Despite the higher sampling rate of the more recent campaigns, ambiguities remain as to whether the measured Hβ lags represent the true mean BLR radius, as the light curves were missing data at key points in time. We see in our velocity-delay map that the majority of the response in the Hβ emission seems to be centered on a delay of 30 days (Figures 5.12 and 5.13). ∼
To investigate this, we ran a one-dimensional delay map analysis of PG2130+099 in MEMECHO to look for an indication of where the true lag lies. Figure 5.15 shows the model continuum light curve envelope in the bottom panel, and the Hβ light curve from Grier et al. (2012b) in the top right panel; the top left panel shows the
99 delay map recovered by MEMECHO. The MEMECHO model fits the data fairly well in this case, and there are two clear peaks in the delay map. The stronger peak is centered around 12.5 days, and the slightly weaker peak is centered at 31 days.
We compare this with the two-dimensional velocity-delay map (Figures 5.8, 5.12, and 5.13), which shows a large signal on the blue side of the emission concentrated at around 30 days and a fainter signal to the redward side that stretches down to shorter lags. A plausible model reproducing these results is a nearly face-on disk with the emitting gas located at around 30 light-days, combined with a strong inflowing gas component not necessarily within the plane of the disk. Including an inflow signature when we measure the flux of the entire Hβ emission line could cause our result to be skewed towards shorter mean lags, when the true distance of the virialized gas is closer to 30 days. Because of the lower quality and coarser ∼ sampling of the light curves for this object, we will likely be unable to model this structure in much more detail. However, it is clear from the delay map that the majority of the Hβ signal comes from a radius of 31 light-days. This radius puts ∼
PG2130+099 much closer to the RBLR–L relation. This also increases the black hole mass estimate for PG2130+099 by a factor of about 2.4, putting it at about 108
M⊙. This would place PG 2130+099 definitely within the expected scatter of the
MBH–σ∗ relation.
100 5.4.2. MEMECHO Parameters and Settings
The A and W Parameters
MEMECHO gives the user control over various aspects of the fitting process.
We first consider the weights wi (see Equation 5.2), which are implemented in
MEMECHO as the user-controlled W and A parameters. The W parameter controls the weight given to pixels in Ψ(λ, τ) relative to the weight given to the pixels in the continuum model C(t). Increasing W makes the delay map stiffer and allows for more flexibility in the continuum model. A affects the aspect ratio of features in the velocity-delay map Ψ(λ, τ). The default values for Ψ(λ, τ) are geometric means of neighboring pixels, with A increasing the weight of neighbor pixels in the λ direction relative to those in the τ direction.
It is important that the structures we see in our delay maps are not dependent on our choices of A and W . To verify that our results are robust, we produced
MEMECHO models altering the A and W parameters and changing the wavelength and velocity grids of the transfer function models. The panels of Figure 5.16 show the resulting velocity-delay maps for 3C 120 as we vary these parameters. For the most part, the general shape of our velocity-delay map is not affected by changes in
A or W . However, if A is made too large, we start to over-fit the data and introduce spurious features into the maps. The stability with respect to the W parameter is expected, since the continuum light curve we are using is very highly sampled,
101 and therefore very highly constrained. Since these behaviors are generic to all four objects, we only show the results for 3C 120.
The Minimum χ2/N
We also investigated the effect of varying the target χ2/N of the MEMECHO solution, as the final resolution of the recovered delay maps is controlled by this parameter. The choice of the target χ2/N is a trade off between smoother, lower delay resolution maps for higher χ2/N, and sharper but less reliable structure in the maps at lower χ2/N. Because we are using over-sampled continuum model light curves from JAVELIN, the actual value of χ2/N is no longer strictly valid, because the continuum light curve data points are not fully independent. Therefore, our chosen values of χ2/N reflect the best trade-off between delay-map resolution and reliable structure in the fits that we could obtain. Figure 5.17 shows velocity-delay maps for 3C120 as we vary the χ2/N, from left to right. As we lower the target
χ2/N, the structure eventually becomes more complex but also less reliable, as the fits are now producing structures to model the noise in the line light curves.
However, the basic structure of the velocity-delay maps is robust to reasonable changes in the target χ2/N.
102 Minimum Allowed Time Delays
When we vary the minimum lag τmin that we allow MEMECHO to consider, we also see a trend worth noting. Figure 5.17 shows that when τmin is set to zero
(top panels), the velocity-delay structure in the map also extends all the way to zero. However, when negative lags are allowed (middle and bottom panels), we see that the structure in the velocity-delay maps, particularly when considering the chevron-shaped structure in the Hβ and Hγ emission lines, peaks at a positive delay and is lower at τ = 0 but does not extend to 0. This has to do with the way MEMECHO deals with its delay map models. The default is set to “pull” the response down to zero at the ends of the delay map when τmin and τmax are not equal to zero. Therefore, when τmin = 0, the default cannot pull down at the low-τ end, so the lowest-entropy map is an exponential function of τ. With τmin < 0, the entropy pulls down on both ends of the delay map, which then creates a Gaussian peak at positive delays.
Because of this, different values of τmin can affect the final velocity-delay maps produced, as this effectively regulates the behavior of the model on some level when trying to smooth the delay maps to the default image. We currently have no means to evaluate which value of τmin results in the “correct” velocity-delay map — it is therefore worth considering velocity-delay maps with various τmin parameters, as we do here, when drawing conclusions about the BLR structure signatures, and any
103 detailed modeling and interpretations should consider and evaluate the differences in these maps. In this work, we make only qualitative statements regarding the signatures seen in the velocity-delay maps. Whether or not the response actually reaches zero at any point in the 3C 120 maps is unknown, but for our purposes it is reassuring that all of the maps show the same basic structures and kinematic signatures. From Figure 5.17, comparing results for τ = 10, 1, and 0, we can min − − also evaluate the evidence for a deficit of prompt response from the resulting delay maps created with each value of τmin. At all three values of τmin, we see Gaussian delay distributions at positive lags in the core of Hβ and Hγ in 3C120, but not for
He ii — thus the data provide evidence for a deficit of prompt response in the center of Hβ and Hγ, but not for He ii. This behavior is consistent with our expectations for the stratification of the BLR.
104 Fig. 5.1.— Velocity-binned reverberation lag results. The top panels show the rms residual spectrum for each object, with the edges of the bins designated by vertical dashed lines. The bottom panels show the mean time delays measured for each bin. The zero-velocity center of the Hβ emission line is shown by the dotted line. The horizontal solid line shows the average time lag reported in Grier et al. (2012b), with uncertainties as horizontal dashed lines. Error bars in the velocity direction show the width of the bins. 105 10 9 8 7 6 2 1.8 1.6 1.4
4.5 4 3.5 3 12
10
8
3.5
3
5450 5500 5550
Fig. 5.2.— Original continuum light curves (gray data points) from Grier et al. (2012b) and the JAVELIN mean continuum model (black solid line) used in the MEMECHO analysis. The dotted line shows the standard deviation about the mean light curve from individual model realizations. Fluxes are given in units of 10−15 erg s−1 cm−2A˚−1.
106 Fig. 5.3.— Best MEMECHO fits to the spectra of Mrk 335. The left panels show the delay maps at each selected rest-frame wavelength, given in the top righthand corner of each panel. The right panels show the emission-line light curve at these selected wavelengths and the MEMECHO fit to the light curve. The χ2/N of the fit for each light curve is shown in each panel. The bottom panel shows the original continuum light curve (black error bars), the error envelope from the simulated light curve showing the standard deviation about the mean (gray envelope), and the continuum model from MEMECHO (solid black line).
107 Fig. 5.4.— Best MEMECHO fits to the spectra of Mrk 1501. Panels and symbols are the same as in Figure 5.3.
108 Fig. 5.5.— Best MEMECHO fits to the spectra of 3C 120. Panels and symbols are the same as in Figure 5.3.
109 Fig. 5.6.— Best MEMECHO fits to the spectra of Mrk 6. Panels and symbols are the same as in Figure 5.3.
110 Fig. 5.7.— Best MEMECHO fits to the spectra of PG 2130+099. Panels and symbols are the same as in Figure 5.3.
111 Fig. 5.8.— Our best-fit velocity-delay maps over the full rest-frame wavelength range for each object (grayscale), projections onto the rest-frame wavelength axis (bottom panel) and the time-delay axis (right panel). Ψ(λ) is the overall response added up at each wavelength, and Ψ(τ) is the overall response of all emission lines added together at a given τ.
112 Fig. 5.9.— Velocity-delay maps for both emission lines seen in Mrk 335. The dotted 2 6 lines show the “virial envelope”, V τc/G = 4.6 10 M⊙, measured from the mean time lag (Grier et al. 2012b). ×
113 Fig. 5.10.— Velocity-delay maps for both emission lines seen in Mrk 1501. The dotted 2 7 lines show the “virial envelope”, V τc/G = 3.3 10 M⊙, measured from the mean × time lag (Grier et al. 2012b).
114 Fig. 5.11.— Velocity-delay maps for all three emission lines seen in 3C 120. The 2 7 dotted lines show the “virial envelope”, V τc/G =1.2 10 M⊙, measured from the × mean time lag (Grier et al. 2012b). The velocity-delay map for Hγ is truncated at V 3250 km/s because the wavelength range of our spectrograph did not cover this far∼ into the red.
115 Fig. 5.12.— Velocity-delay maps for all three emission lines seen in PG 2130+099. 2 6 The dotted lines show the “virial envelope”, V τc/G =8.3 10 M⊙, measured from × the mean time lag (Grier et al. 2012b).
116 Fig. 5.13.— False-color velocity-delay maps for Mrk 335, Mrk 1501, 3C120, and PG 2130+099. The dotted lines in each panel correspond to virial envelopes for each object as listed in Figures 5.9-5.12. The Hβ emission is shown in red, Hγ emission in green, and He ii λ4686 emission in blue.
117 Fig. 5.14.— Velocity-delay maps for simple BLR models of Hβ emission around a 7 1 10 M⊙ black hole. Panel (a) shows highly beamed emission (typical for Hβ, see × Ferland et al. 1992) from gas in free-fall motion. The infalling gas is distributed in a spherical shell with inner and outer radii of 15 and 20 light-days, inclined at an angle of 45 degrees. Panel (b) shows the same infall model but with a more-extended BLR with inner and outer radii of 5 and 20 light-days, respectively. Panel (c) shows a map for an edge-on Keplerian disk with inner and outer radii of 5 and 20 light-days, and panel (d) shows a map for a fully illuminated thick spherical shell of Keplerian circular orbits, with inner and outer BLR radii of 5 and 20 light-days.
118 Fig. 5.15.— One-dimensional delay map for PG 2130+099. The bottom panel shows the simulated continuum light curve used in the MEMECHO analysis, with the errors shown as the black envelope. The top right panel shows the light curve for the entire Hβ emission line from Grier et al. (2012b). The top left panel shows the one-dimensional delay map from MEMECHO.
119 Fig. 5.16.— Velocity-delay maps for 3C 120. Each panel shows the best map found as we vary A and W by factors of 100. Increasing A smooths the maps more strongly in λ and less in τ, while increasing W reduces the importance of fitting the continuum model relative to the delay maps. All models are converged to the same overall goodness-of-fit χ2/N. Our adopted velocity-delay map is that of the middle panel.
120 Fig. 5.17.— Velocity-delay maps for 3C 120, varying the degree of smoothing and the 2 minimum lag allowed in the model (τmin). Increasing values of χ /N mean that the entropy term is more heavily smoothing the map. Our adopted velocity-delay map is that of the top middle panel.
121 2 Object A W τmin χ /N
Mrk335 0.1 1.0 -1 2.8 Mrk1501 1.0 1.0 0 1.8 3C120 1.0 1.0 0 2.4 PG2130+099 0.1 1.0 0 2.1
Note. — Parameters are defined in Section 5.3 and 5.4.2.
Table 5.1. MEMECHO Parameters
122 Chapter 6
Stellar Velocity Dispersion Measurements in High-Luminosity Quasar Hosts and Implications for the AGN Black Hole Mass Scale
6.1. Background
An important step in evaluating the MBH–σ∗ relation and any possible deviations from it is to obtain secure σ∗ and MBH measurements in AGNs that sample the entire mass range of the relation. While the high-mass end of the
9 quiescent MBH–σ∗ relation is relatively well-populated to beyond 10 M⊙ (McConnell
& Ma 2013), the current sample of AGNs used to calculate f still contains just h i 8 three or four objects with MBH above 10 M⊙ (Woo et al. 2010; Graham et al. 2011;
Park et al. 2012). More measurements for luminous AGNs are needed to measure the high-mass end of the AGN MBH–σ∗ relation. However, accurate σ∗ measurements for high-luminosity AGNs are difficult to obtain because the AGN light overpowers the light from the host. Moreover, more luminous AGNs are relatively scarce and thus typically found at large distances, so the host galaxy has a small angular size and is easily lost in the glare of the AGN. It is only in the past few years
123 that high-precision measurements in very luminous objects have been obtained on account of the availability of adaptive optics (AO) and integral field spectrographs
(IFUs) such as Gemini North’s Near-Infrared Integral Field Spectrometer (NIFS) combined with the Gemini North laser guide star AO system, ALTAIR. Watson et al. (2008) used NIFS+ALTAIR to measure σ∗ for PG 1426+015 with much higher precision than previous measurements for high-luminosity quasars. This success prompted us to undertake additional observations of quasars at the high-mass end of the MBH–σ∗ relation. In this paper we present the results of our NIFS observations of eight high-luminosity quasars. We successfully measured σ∗ in four objects and use these results to improve the population of the MBH–σ∗ relation at the high-mass end. We also recalculate virial products for the entire AGN sample with updated time lag measurements to re-derive f , calibrate black hole masses in AGNs, and h i reexamine the AGN MBH–σ∗ relation. In this section we again adopt a cosmological
−1 −1 model of Ωm =0.3, ΩΛ =0.70, and H0 = 70 km sec Mpc .
6.2. OBSERVATIONS AND DATA ANALYSIS
6.2.1. NIFS/ALTAIR Observations
Observations of eight quasars were carried out at the Gemini North telescope in 2008 and 2010 under the programs GN-2008B-Q-28, GN-2010A-Q-11, and
GN-2010B-Q-24. We chose our sample from the database of objects with
124 reverberation-based black hole mass measurements from Peterson et al. (2004) with
8 MBH > 10 M⊙. Basic information on our targets is given in Table 6.1. We used
NIFS in conjunction with the ALTAIR laser guide star AO system to carry out our observations. NIFS has a 3′′ 3′′ field of view that is divided into 29 individual × spectroscopic slices, with a spectral resolution R = λ/∆λ 5290 in both the H and ≈ K bands. With the AO correction, NIFS yields a spatial resolution on the order of 0.1′′. Seven of our targets were observed in the H band, which has a central wavelength of 1.65 µm and covers from about 1.49 µm to 1.80 µm. We observed our eighth object, PG1700+518, in the K band due to its higher redshift. The K filter on NIFS covers from about 1.99 µm to 2.40 µm.
We estimated the integration time for each object with HST ACS or WFPC2 images of the sources from Bentz et al. 2009a. To simulate the data we would obtain from NIFS, we measured the flux within a 3′′x 3′′ aperture, except for a central circle of diameter 0.2′′. We esimated the exposure time for each object based on its brightness relative to PG 1426+015, for which Watson et al. (2008) obtained a host-galaxy signal-to-noise ratio (S/N) 200 in about two hours of on-source ∼ integration. With both Poisson and background-limited trials, we estimated the integration time required for each object to obtain a S/N 200. Table 6.2 gives ∼ details of the observations, most notably the on-source integration time for each object. Reconstructed images from the IFU spectra of all eight targets are shown in
125 Figure 6.1. We observed telluric standard stars (usually A0V) once every 1.5 hours for the purpose of telluric corrections.
6.2.2. Data Reduction
Data were processed through the standard NIFS pipeline1 from the Gemini
IRAF2 package. Our reductions deviated from the standard pipeline tasks in only two ways. First, we found that the original sky frames did not adequately remove sky lines from our spectra. To remedy this, we manually scaled the individual sky spectra to obtain better sky subtraction in each individual object frame. Second, to remove stellar absorption lines in our telluric spectra, we used methods described by Vacca et al. (2003) and applied in the IDL-based code xtellcor. This code uses a theoretical model of Vega to remove the hydrogen features in our telluric standard spectra, and is specifically written for use with A0V stars.
To separate the host-galaxy spectra from the AGNs, we extract the spectrum from an annulus that excludes the quasar-dominated nucleus. The use of AO in these observations allowed us in most cases to confine at least the core of the quasar
flux to the very central pixels of the image. Generally, the AO-assisted seeing was on
1http://www.gemini.edu/sciops/instruments/nifs 2IRAF (Tody 1986) is distributed by the National Optical Astronomy Observatories, which are operated by the Association of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation.
126 the order of 0.1 0.2′′, so we used either a 0.2′′, 0.3′′, or 0.4′′ inner radius (R ) to − inner isolate the quasar component. The outer radius for each extraction annulus (Router) was chosen to include as much host galaxy light as possible while minimizing the amount of noise contributed by the sky. For most of our objects, we evaluated this by eye and chose windows that minimized noise. However, in the targets where we could see identifiable galactic absorption lines, we chose Rinner and Router to obtain the highest equivalent width measurements in the visible absorption lines. Rinner and Router for each object are listed in Table 6.3 and are shown on the reconstructed images in Figure 6.1. The total galaxy+quasar spectra for all eight objects, extracted from within a radius of Router, are shown in the top panels of Figures 6.2, 6.3, 6.4, and 6.5.
Although the use of the ALTAIR AO system helps confine the nuclear light to the central few pixels of the image, the AO-corrected PSF still has nuclear light in its wings. Typical Strehl ratios for the ALTAIR AO system are on the order of
0.1 to 0.3 in the H band (Christou et al. 2010). For a typical Strehl ratio of 0.2,
Christou et al. (2010) report a 50% encircled PSF energy radius of about 0.4′′; i.e., half of the quasar light falls in a radius outside 0.4′′. Thus, there is still significant quasar contamination in the host-galaxy spectra. To further remove the quasar emission, we scaled and subtracted the nuclear spectrum from our annulus in each target. This eliminated much of the quasar emission from our spectra. Our final nucleus-subtracted, observed-frame spectra of all eight objects are shown in the
127 bottom panels of Figures 6.2, 6.3, 6.4, and 6.5. Note that while the spectrograph coverage in the H band extends from about 1.48 µm to 1.8 µm, in some cases the telluric contamination was sufficiently significant that we cropped the spectra before using them. The wavelength coverage of the H-band spectra shown in Figures
6.2–6.5 is not uniform for this reason.
6.2.3. Stellar Velocity Dispersion Measurements
We used the penalized pixel fitting method (pPXF) of Cappellari & Emsellem
(2004) to measure σ∗. This method convolves a stellar template spectrum and a line-of-sight velocity distribution to model the host galaxy. The best line-of-sight velocity distribution is calculated in the pPXF code with a χ2 minimization technique. The velocity templates used to make our measurements were obtained by Watson et al. (2008), and include stars of four different spectral classes: K0 III,
K5 III, M1 III, and M5 Ia. We list the most prominent stellar absorption features in these spectra in Table 6.4. The K5 III, M1 III, and M5 Ia templates all resulted in somewhat similar fits in each spectrum — the reduced χ2 values of the fit in these three cases were always very close to one another, with the K5 III template usually a slightly better fit than the other two. The K0 III template provided a poor fit to the
CO(3-0) absorption line in all objects and was therefore not used in our analysis.
128 There are several different factors that affect the uncertainty in σ∗. First, no single stellar template is expected to be a perfect match to the host-galaxy stellar absorption features. We therefore adopt the average σ∗ value from the three templates as our estimate, and fold the differences into our uncertainties. One exception to this is Mrk 509, for which the K5 III template provided a significantly better fit. We take the standard deviation in σ∗ reported by the pPXF software among the three stellar templates as representative of the template mismatch uncertainty. We also consider the location of ∆χ2 = 1 in the fitting a component of our uncertainties. To do this, we allowed σ∗ to vary but held the rest of the parameters fixed at their best-fit values. We identified the value of σ∗ at which the
χ2 had changed by 1 from the best-fit value, and took the average difference ± between these two values and our best-fit value as our uncertainty. Because our measurements for PG 1411+442 and PG 1617+175 were made with spectra that are significantly noisier than the other two spectra, we also include a signal-to-noise
(S/N) component in our uncertainties for those two objects. To estimate this component, we degraded our two best spectra (Mrk 509 and PG 2130+099) to match the S/N in PG1617+175 and PG1411+442. We then re-measured σ∗ in the degraded spectra and took the deviation from our original measurements to represent the S/N component of the uncertainties. For most of the templates, the
σ∗ measurements tended to be overestimated in the degraded spectra by about 20
129 km s−1. We combine the 1σ uncertainties, the template mismatch uncertainties, and any S/N component in quadrature and adopt these as our formal uncertainties.
We measured σ∗ from the spectra of four of our objects: Mrk 509, PG1411+442,
PG1617+175, and PG2130+099, and these values are listed in Table 6.3. The normalized galaxy spectra and the best-fit broadened stellar templates for these four objects are shown in Figure 6.6 and Figure 6.7. Our measurement for PG2130+099 of 147 17 km s−1 constitutes a significant improvement in precision over its ± previous measurement of 172 46 km s−1 (Dasyra et al. 2007), while Mrk 509 ± and PG 1411+442 have no previous measurements reported in the literature. Our measurement for PG1617+175 of 201 37 km−1 is slightly more precise but ± consistent with measurements made by Dasyra et al. (2007), who report σ∗ = 183
47 km s−1. In the other four objects, we were unable to identify any absorption ± lines in our nucleus-subtracted spectra and were unable to fit stellar templates and recover σ∗. We believe this was caused mainly by the overwhelming strength of the quasar emission in these objects and a lack of strong absorption lines within the observed wavelength range. In an attempt to minimize the quasar contamination, we experimented with different radii for both the inner and outer regions in these objects, but in all cases were unable to remove the quasar contamination enough to see absorption in the host galaxy spectrum. Even when subtracting off a scaled nuclear spectrum to eliminate the emission lines, we see only noise (with residual
130 sky and telluric contamination) in the spectra of objects in which we were unable to measure σ∗ (see Figures 6.2–6.5).
Because we optimized our extraction radii to obtain the best host-galaxy-to- quasar ratio, our σ∗ measurements were made within different effective physical apertures for each target. The measurements for Mrk 509 and PG1411+442, and
PG 1617+175 were made within an effective physical aperture of 1.25 kpc, 3.2 kpc, and 3.3 kpc respectively. The effective radii (re) for the spheroid components of these three objects from the surface decompositions of Bentz et al. (2009a) are 1.85 kpc, 5.05 kpc, and 3.3 kpc, so our apertures for these three objects lie between, or very close to, the commonly quoted aperture sizes of re and re/8. However, the measurement for PG 2130+099 was made within 1.64 kpc, which is 4.3 times re for this object. We apply the relation derived by Jorgensen et al. (1995) to determine the velocity dispersion within the effective radius (σ∗,e) for all four objects. These values are listed in Table 6.3 for comparison with our measurements. For Mrk
509, PG1411+442, and PG1617+175, there is very little difference between our measured σ∗ and σ∗,e. The value of σ∗,e for PG2130+099 is notably higher than our quoted value, but the change does not qualitatively affect our analysis. We use the corrected σ∗ measurements (σ∗,e) for the subsequent discussion and analysis.
131 6.2.4. Recalculation of AGN Virial Products
There are now 30 reverberation-mapped AGNs with σ∗ measurements. The virial products for most of these objects were calculated from time lags that were determined with traditional cross correlation methods (e.g., Peterson et al. 2004;
Bentz et al. 2009c). However, Zu et al. (2011) recently introduced a different method to determine time lags that they called Stochastic Process Estimation for AGN
Reverberation (SPEAR). This method works as follows: We assume all emission-line light curves are scaled and shifted versions of the continuum light curve. The continuum is modeled as an autoregressive process using a damped random walk
(DRW) model, which has shown to be a good statistical model of quasar variability
(Gaskell & Peterson 1987; Kelly et al. 2009; Koz lowski et al. 2010; MacLeod et al.
2010; Zu et al. 2013). The transfer function is modeled as a simple top-hat function.
We fit the continuum and emission-line light curves simultaneously, maximizing the likelihood of the model using Bayesian Markov Chain Monte Carlo iterations.
The main advantage of the SPEAR method is that it treats gaps in the temporal coverage of light curves in a statistically self-consistent way, so the gaps in the data are filled in a well-defined manner with well-defined uncertainties.
Zu et al. (2011) re-measured the time lags in reverberation-mapped AGNs with the SPEAR method and demonstrated its ability to recover accurate time lags. The method has since been successfully used to improve reverberation measurements
132 (Grier et al. 2012b; Dietrich et al. 2012) and even recover velocity-delay maps
(Grier et al. 2013). Because many of the light curves from the AGN MBH–σ∗ sample, particularly high-luminosity objects, have large gaps, we went through and re-determined the virial products for the entire sample with updated Hβ time lags recovered with the SPEAR method. Eighteen of the objects already have Hβ time lags from Zu et al. (2011). For eight of the remaining objects, we applied the same SPEAR method with the latest version of the software called JAVELIN3 to calculate new time lags and virial products. Three of the remaining AGNs have recently-published virial products calculated with the SPEAR method, so we use the published virial products for 3C 390.3 from Dietrich et al. (2012), 3C120 from Grier et al. (2012b), and PG2130+099 from Grier et al. (2013). We do not have the light curve for Mrk 50, so for this object we use the virial product from Barth et al. (2011).
All recalculated time lags, original line widths, and updated virial products for each reverberation data set are given in Table 6.5, 6.6, and 6.7. For objects with just one measurement, we use that virial product. For objects with multiple measurements, we adopt the mean of the logarithm of the virial products. In Table 6.8 we show the adopted virial products and σ∗ measurements for the entire sample. In most cases, the updated virial products are very similar to the previously-quoted values.
However, for a few objects with sparsely sampled light curves, the virial products changed noticeably from prior calculations.
3Available at: http://www.astronomy.ohio-state.edu/ yingzu/codes.html#javelin ∼ 133 6.3. Results and Discussion
6.3.1. Data Quality
The objects for which we successfully measure σ∗ (PG 1411+442, PG 1617+175,
Mrk 509, and PG2130+099) are the four lowest-redshift galaxies in our sample. Mrk
509 and PG 2130+099 have the highest S/N host-galaxy spectra, with an average
S/N per pixel of 250 for Mrk 509 and 190 for PG 2130+099. These have the ∼ ∼ most easily identifiable galaxy absorption features in our sample (see Figures 6.6 and 6.7). The higher-luminosity quasars, PG 1411+442 and PG 1617+175, have lower S/N, specifically a S/N of 100 per pixel in PG 1617+175 and 130 per pixel in PG 1411+442, which is lower than our anticipated S/N. Although we made several attempts to fit and remove sky features, residual sky contamination and telluric absorption lines remain in the subtracted spectra, which makes the velocity dispersion measurements more uncertain and contributes to the lower S/N of the spectra. We also see stronger quasar emission features which we were unable to eliminate entirely from the host-galaxy spectra.
There seems to be three main factors that compromise the quality of our host galaxy spectra. First, we were unable to satisfactorily remove the sky emission from the spectra in all eight targets. This caused a significant decrease in S/N in all of our spectra. Secondly, in three of our objects, PG 0026+129, PG1126+023,
134 and PG0052+251, many of the strong stellar absorption features that allow us to measure σ∗ were redshifted out of the H-band. This limited us to very few absorption lines, and these few remaining lines fell in regions with severe telluric contamination. As such, we did not detect any absorption and were unable to measure σ∗ in these objects. Third, and possibly most importantly, in these four objects the quasar contamination becomes strong enough to overwhelm the host galaxy flux despite our long integrations and attempts to optimize the extraction radius. To quantify the amount of quasar contamination remaining in the original extraction annuli of these spectra, we estimated the ratio of the quasar flux to the host flux, both within our extraction radius, for the case of PG0026+129. We based this calculation on measurements of the PSF magnitude, host galaxy magnitude, and host galaxy Sersic index reported by Veilleux et al. (2009) from their analysis of
HST NICMOS H-band images. We integrate the Sersic profile over the extraction annulus used in our study (for PG 0026+129, we used an inner radius of 0.2′′and an outer radius of 0.6′′, corresponding to 0.62 and 1.86 kpc, respectively) to estimate the amount of host flux inside our aperture. Given the previously discussed findings of
Christou et al. (2010), we assume that half of the PSF light falls inside the annulus and find that the PSF flux inside the extraction annulus is a factor of 4.5 times the amount of host flux inside the extraction annulus. We expect similar, possibly even more, contamination in the other targets for which we were unsuccessful, and thus
135 this contamination limits our ability to explore the hosts of quasars at the high end of the luminosity distribution.
While we appear to have reached the limit of the NIFS+ALTAIR system for these measurements, AO systems continue to move towards diffraction-limited resolution with high Strehl ratios. These advances may lead to successful measurements with similar exposure times. The future availability of the
James W ebb Space T elescope (JWST ) may also lead to successful attempts at σ∗ measurements in high-luminosity quasars. JWST is currently expected to launch in 2018 and will be equipped with an IFU spectrograph of sufficient resolution, and the major problems of sky and telluric contamination, which were prohibitive for our higher-redshift targets, will be completely eliminated in space. Observing from space will allow us to see the whole spectrum continuously, so we will not be limited to specific redshift windows, and will also include the Ca ii triplet region.
JWST will offer a compact, stable PSF, and will have more sensitivity than our current equipment and thus offers much promise for future efforts to measure σ∗ in high-luminosity, high-redshift AGNs.
6.3.2. The Virial Factor f and the MBH–σ Relation h i ∗
Because we have completely updated the virial products in the AGN MBH–σ∗ sample and added a few objects at the high-luminosity end of the distribution,
136 we also present an updated measurement of the average virial factor f used to h i calibrate the AGN M scale. In order to measure f , we assume that the AGN BH h i
MBH–σ∗ relation follows the same slope as quiescent galaxies. However, previous studies have found that AGNs appear to follow a slightly shallower relation than quiescent galaxies (e.g., Woo et al. 2010; Graham et al. 2011; Park et al. 2012). We measure the slope of the relation between the virial product, Mvir, and σ∗:
σ∗ log M = α + β log (6.1) vir 200 kms−1 with our updated AGN sample. We use the traditional forward regression for our calculation: We consider σ∗ as the independent variable and MBH the dependent variable. Using the FITEXY algorithm (Press et al. 1992) to fit the relation, we obtain β = 5.01 0.20. This is steeper than found in previous studies (e.g., Woo ± +0.40 et al. 2010, who report β = 3.72−0.39), and is very close to that measured in quiescent galaxies by Park et al. (2012) with the McConnell et al. (2011) galaxy sample
(β = 5.07 0.36, using the FITEXY estimator and forward regression approach). ± Following Woo et al. (2010) and Park et al. (2012), we then use the FITEXY estimator to determine the f necessary to place the AGN sample on the same h i
MBH–σ∗ relation as quiescent galaxies. Using our updated AGN sample, we obtain log f = 0.62 0.11, corresponding to f = 4.19 1.08. This number is slightly h i ± h i ± lower than, but consistent with, the recent values found by Woo et al. (2010) and
Park et al. (2012).
137 We use our new value of f = 4.19 to transform the virial products of the h i
AGN sample to MBH. We place our four objects with successful σ∗ measurements on the MBH–σ∗ relation in Figure 6.8 with the rest of the AGN sample shown for comparison. We also show the most recent quiescent MBH–σ∗ relation from Park et al. (2012), with the quiescent galaxy sample from McConnell et al. (2011) with a forward regression calculation. We see that all four of our objects lie within the expected scatter of the relation. There is currently no evidence for an offset in the objects at the high end of the relation. Because we were unable to obtain σ∗ measurements for our highest-MBH objects, we fall short of our original goal of densely populating the very high end of the MBH–σ∗ relation, though we do increase
8 9 the sample of objects with 10 M⊙ < MBH < 10 M⊙. As noted by Woo et al.
(2010) and Park et al. (2012), the AGN sample is still biased towards objects with relatively lower masses than the majority of the quiescent galaxies with dynamical mass measurements.
A morphological or environmental dependence of the MBH–σ∗ relation could have important implications on the use of a single mean f factor to transform the virial product, Mvir, into MBH in AGNs. To test whether or not the AGN sample shows these morphology-dependent effects, we divided our AGN sample into different groups to see if there was any visible offset or change in slope in the
MBH–σ∗ relation. In the left panel of Figure 6.9, we divide the AGN sample into two groups: those suspected of hosting a pseudobulge, and those thought to host a
138 classical spheroid at their centers. In the right panel, we divide the sample based on whether or not a central bar is observed. We classified the galaxies using the host-galaxy decompositions from Bentz et al. (2009a) (see their Table 4). Following
Kormendy & Kennicutt (2004), the galaxies were inspected for the presence of the following pseudobulge indicators: flattened bulge morphology, the presence of a nuclear bar, a low (n < 2) Sersic index, and the presence of copious dust and star formation in the nucleus without any signature of an ongoing merger. None of these indicators by themselves constitute sufficient evidence that a bulge is a pseudobulge. Therefore, we assumed that galaxies with most of these indicators host a pseudobulge, and galaxies with few or none of these indicators host a classical bulge. Our classifications are listed in Table 6.8. It should be noted that host galaxy classifications are particularly difficult to make in high-luminosity quasar hosts, and some galaxies in our sample show signs of disturbed morphologies, so our classifications are somewhat uncertain. However, we do not see any signs of an offset in either the galaxies hosting pseudobulges or the galaxies with central bars.
If we determine f with our samples of barred and unbarred galaxies separately, h i we obtain a result similar to that of Graham et al. (2011). Namely, the galaxies with bars in them yield a lower f (3.31 1.10) than the galaxies without central bars h i ± ( f = 5.14 1.59). However, the barred sample lacks any objects with M > 108 h i ± BH 7 M⊙, and similarly, the unbarred sample has only three objects with MBH below 10
M⊙. As demonstrated by Woo et al. (2010) and this work, extending the sample to
139 cover the entire MBH range can significantly impact the measured slopes and mean virial factor, and as such we hesitate to attribute meaning to virial factors calculated with samples lacking a dynamic range in MBH.
6.4. Summary
We measured σ∗ in four quasars at the high-mass end of the AGN MBH–σ∗ distribution. The measurements of Mrk 509 and PG 1411+442 are the first ever
σ∗ measurements for these targets, and the measurements for PG 1617+175 and
PG 2130+099 are updates with improved precision. We were unable to measure
σ∗ in the spectra of our highest-redshift objects due to substantial contamination of the host-galaxy spectra by the quasar nucleus and poor sky line subtraction.
Future measurements of σ∗ for high-luminosity QSOs will require improved AO technology, substantially longer integration times, or JWST . We also updated the virial products for the AGN MBH–σ∗ sample with recalculated time lags from the
SPEAR method of Zu et al. (2011) and recalculated the mean virial factor f used h i to calibrate the AGN M scale. We obtained log f = 0.62 0.11, corresponding BH h i ± to f = 4.19 1.08. This is consistent with previous results based on forward h i ± regression methods. With our new σ∗ measurements, all four of our objects fall within the expected scatter of the quiescent MBH–σ∗ relation. We find no evidence
140 in AGNs for a morphology-based deviation from the standard quiescent MBH–σ∗ relation.
141 Fig. 6.1.— Raw reconstructed images for each object. The white circles denote the extraction annuli used for each object, given in Table 6.3. The field of view of each panel is 3′′ 3′′. ×
142 Fig. 6.2.— Observed-frame NIFS spectra of PG 0026+129 and PG 0052+251. The top panels show the total, unsubtracted spectrum, and the bottom panels show the host galaxy spectrum after the nucleus was subtracted off. The fluxes are in units of flux per unit wavelength, and have been normalized to the mean of the unsubtracted spectrum for each object.
143 Fig. 6.3.— Observed-frame NIFS spectra of PG 1226+023 and PG 1411+442. See Figure 6.2 for details.
144 Fig. 6.4.— Observed-frame NIFS spectra of PG 1617+175 and PG 1700+518. See Figure 6.2 for details.
145 Fig. 6.5.— Observed-frame NIFS spectra of Mrk 509 and PG 2130+099. See Figure 6.2 for details.
146 1.04 Mg I Mg I CO(3-0) Si I CO (5-2) CO(4-1) CO (6-3) 1.02
1
0.98
0.96
1.5 1.55 1.6
Mg I CO(3-0) 1.04 Mg I CO(4-1)
1.02
1
0.98
0.96 1.5 1.52 1.54 1.56 1.58
Fig. 6.6.— Normalized rest-frame spectra of PG 1411+442 and PG1617+175, in which we measured σ∗ successfully. The red lines are our best-fit models, with the K5 III stellar template for PG 1411+442 and the M5 Ia template for PG1617+175. The shaded gray areas mark areas excluded from the fit due to telluric and/or quasar light contamination.
147 Mg I 1.06 Mg I CO(3-0) 1.04 CO(4-1) Si I 1.02 CO(5-2) CO(6-3) CO(8-5) 1
0.98
0.96 1.5 1.55 1.6 1.65
1.04 Mg I Mg I CO(3-0) CO(4-1) CO(5-2) 1.02 Si I CO(6-3)
1
0.98
1.5 1.55 1.6
Fig. 6.7.— Normalized rest-frame spectra of Mrk 509 and PG 2130+099, in which we measured σ∗ successfully. The red lines are our best-fit models with the K5 III stellar template. The shaded gray areas mark areas excluded from the fit due to telluric and/or quasar light contamination.
148 PG1617
PG1411 PG2130
Mrk 509
Fig. 6.8.— The MBH–σ∗ relation. The gray open squares are AGNs with previous measurements. The sample is composed of the compilation by Woo et al. 2010 and updated by Park et al. 2012, with additional updates as described in the text (see Table 6.8). All MBH were calculated with our measurement of f=4.19. The solid black stars show our new σ∗ measurements. The open circles denote the expected locations of the objects for which we were unable to measure σ∗. The solid black line shows the most recent measurement of the MBH–σ∗ relation in quiescent galaxies from Park et al. (2012) with the McConnell et al. (2011) sample, calculated with a forward regression analysis. Dotted lines show the intrinsic scatter of the quiescent galaxies measured by Park et al. (2012).
149 Fig. 6.9.— The MBH-σ∗ relation for the same updated AGN sample shown in Figure 6.8. The left panel shows the AGN sample divided by the type of bulge (pseudobulge or classical), and the right panel shows the AGN sample sorted by the presence of a bar. All MBH were calculated with our measurement of f=4.19. The solid black line shows the most recent measurement of the MBH–σ∗ relation in quiescent galaxies from Park et al. (2012) with the McConnell et al. (2011) sample using a forward regression analysis. Dotted lines show the intrinsic scatter of the quiescent galaxies measured by Park et al. (2012).
150 RA DEC z Galaxy (J2000) (J2000) (NED)
PG0026+129 00 29 13.6 +13 16 03 0.142 PG0052+251 00 54 25.1 +25 25 38 0.154 PG1226+023 12 29 06.7 +02 03 09 0.158 PG1411+442 14 13 48.3 +44 00 14 0.089 PG1617+175 16 20 11.3 +17 24 28 0.112 PG1700+518 17 01 24.8 +51 49 20 0.292 Mrk509 204409.7 -104325 0.034 PG2130+099 21 32 27.8 +10 08 19 0.063
Table 6.1. Quasar Properties
Observing On-source Target Semester Integration time Band (hours)
PG0026+129 2008B 1.33 H PG0052+251 2010B 1.33 H PG1226+023 2010A 0.83 H PG1411+442 2010A 1.33 H PG1617+175 2010A 2.50 H PG1700+518 2010A 3.00 K Mrk509 2008B 2.00 H PG2130+099 2010B 1.00 H
Table 6.2. Gemini NIFS Observations
151 Galaxy Rinner Router Stellar σ∗ Best-fit σ∗,e (arcsec) (arcsec) Template (km s−1) χ2 DOF χ2/DOF (km s−1)
PG0026+129 0.2 0.6 ··· ··· ··· ··· ··· ··· PG0052+251 0.2 0.7 ··· ··· ··· ··· ··· ··· PG1226+023 0.4 1.4 ··· ··· ··· ··· ··· ··· PG1411+442 0.2 1.6 K5III 216 31 858.80 914 0.940 209 30 ± ± PG1617+175 0.3 1.3 M5Ia 201 37 491.33 649 0.757 201 37 ± ± PG1700+518 0.4 1.1 ··· ··· ··· ··· ··· ···
152 Mrk509 0.3 1.7 K5III 189 12 1046.17 1134 0.923 184 12 ± ± PG2130+099 0.2 1.2 K5III 147 17 548.70 914 0.600 163 19 ± ±
Note. — Rinner and Router correspond to the inner and outer radii of the circular extraction annulus for each object. χ2 and χ2/DOF are reported for the best fits with the K5III stellar template for PG1411+442, Mrk 509 and PG2130+099. For the case of PG1617, the M5Ia
template was used. Rinner and Router correspond to the inner and outer radii of the extraction
annulus. σ∗,e was calculated with the formula for E and S0 galaxies from Jorgensen et al. (1995).
Table 6.3. NIFS Extraction Windows and Measurements Feature Rest-frame Wavelength (µm)
MgI 1.4880 MgI 1.5030 CO(3-0) 1.5580 CO(4-1) 1.5780 SiI 1.5890 CO(5-2) 1.5980 CO(6-3) 1.6190 CO(8-5) 1.6610 CO(9-6) 1.6840 CO(10-7) 1.7060
Table 6.4. Most Prominent Stellar Absorption Features
153 a Galaxy τHβ τHβ σline(rms) σline(rms) Mvir −1 6 (days) Reference (km s ) Reference (10 M⊙)
+1.1 +1.2 3C120 27.2−1.1 1 1514 65 1 12.2−1.2 +3.0 ± +36 3C390.3 44.3−3.3 2 5455 278 2 260−23 +6.6 ± +5.6 Ark120 34.7−8.9 3 1959 109 7 25.9−7.3 +3.2 ± +2.4 Ark120 28.8−5.7 3 1884 48 7 19.9−4.1 +0.7 ± +0.2 Arp151 3.6−0.2 4 1252 46 8,9 1.1−0.1 +0.8 ± +0.9 Mrk50 10.4−0.9 5 1740 101 5 6.2−0.9 +2.8 ± +8.2 Mrk79 25.0−14.1 3 2137 375 7 22.2−14.8 +1.4 ± +1.6 Mrk79 30.2−2.1 3 1683 72 7 16.7−1.8 +7.1 ± +4.9 Mrk79 16.8−2.2 3 1854 72 7 11.3−1.7 +1.7 ± +7.8 Mrk79 42.6−0.8 3 1883 246 7 29.5−7.7 +2.2 ± +1.7 Mrk110 24.4−12.7 3 1196 141 7 6.8−3.9 +5.9 ± +2.1 Mrk110 32.7−5.2 3 1115 103 7 7.9−1.9 +2.1 ± +0.3 Mrk110 20.8−2.0 3 755 29 7 2.3−0.3 +0.1 ± +0.1 Mrk202 3.5−0.1 4 659 65 8,9 0.3−0.1 +1.2 ± +1.1 Mrk279 17.8−1.1 3 1420 96 7 7.0−1.0 +0.3 ± +1.0 Mrk509 67.6−0.3 3 1276 28 7 21.5−1.0 +1.8 ± +0.5 Mrk590 18.5−2.5 3 789 74 7 2.3−0.5 +1.9 ± +1.6 Mrk590 19.0−3.9 3 1935 52 7 13.9−2.9 +3.4 ± +1.5 Mrk590 31.8−8.6 3 1251 72 7 9.7−2.8 +2.4 ± +2.0 Mrk590 30.1− 3 1201 130 7 8.5− 2.3 ± 2.0
Note. — Time lags are all given in the rest frame. References: 1. Grier et al. 2012b; 2. Dietrich et al. 2012; 3. Zu et al. 2011; 4. This work; 5. Barth et al. 2011; 6. Grier et al. 2013; 7. Peterson et al. 2004; 8. Bentz et al. 2009b; 9. Park et al. 2012; 10. Denney et al. 2010 11. ?; 12. Denney et al. 2006
Table 6.5. Reverberation Measurements and Updated Virial Products for the First Part of the AGN Sample
154 a Galaxy τHβ τHβ σline(rms) σline(rms) Mvir −1 6 (days) Reference (km s ) Reference (10 M⊙)
+2.2 +1.2 Mrk817 20.3−2.2 3 1392 78 10 7.7−1.2 +1.8 ± +1.9 Mrk817 16.7−2.6 3 1971 96 10 12.6−2.3 +4.7 ± +4.6 Mrk817 34.8−5.6 3 1729 158 10 20.3−5.0 +1.5 ± +4.7 Mrk817 10.5−1.0 3 3150 295 10 20.3−4.2 +0.9 ± +0.2 Mrk1310 4.2−0.1 4 755 138 8,9 0.5−0.2 +6.1 ± +4.5 NGC3227 10.6−6.1 3 1925 124 10 7.6−4.5 +0.3 ± +0.7 NGC3227 4.4−0.5 3 2018 174 10 3.5−0.7 +1.4 ± +0.7 NGC3516 14.5−1.1 3 1591 10 10 7.1−0.6 +0.3 ± +0.7 NGC3783 7.2−0.7 3 1753 141 7 4.3−0.8 +0.1 ± +0.1 NGC4051 2.5−0.1 3 1034 41 10 0.5−0.1 +0.6 ± +0.9 NGC4151 6.0−0.2 3 2680 64 11 8.4−0.5 +0.2 ± +0.2 NGC4253 5.4−0.8 4 516 218 8,9 0.3−0.2 +0.7 ± +0.4 NGC4593 4.5−0.6 3 1561 55 12 2.1−0.3 +0.6 ± +0.2 NGC4748 8.6−0.4 4 657 91 8,9 0.7−0.2 +0.8 ± +0.9 NGC5548 20.8−1.0 3 1687 56 7 11.6−0.9 +0.8 ± +1.1 NGC5548 16.0−1.3 3 1882 83 7 11.1−1.3 +2.1 ± +2.0 NGC5548 15.5−1.1 3 2075 81 7 13.0−1.4 +1.2 ± +1.5 NGC5548 10.8−1.0 3 2264 88 7 10.8−1.3 +1.4 ± +1.7 NGC5548 15.0−2.9 3 1909 129 7 10.7−2.6 +1.4 ± +2.6 NGC5548 10.6−1.0 3 2895 114 7 17.4−2.1 +1.3 ± +3.1 NGC5548 23.8−0.9 3 2247 134 7 23.4−2.9 +0.3 ± +0.9 NGC5548 15.8− 3 2026 68 7 12.7− 0.6 ± 1.0
Note. — See Table 6.5 for details.
Table 6.6. Reverberation Measurements and Updated Virial Products for the Second Part of the AGN Sample
155 a Galaxy τHβ τHβ σline(rms) σline(rms) Mvir −1 6 (days) Reference (km s ) Reference (10 M⊙)
+0.4 +0.8 NGC5548 16.5−0.2 3 1923 62 7 11.9−0.8 +1.5 ± +1.6 NGC5548 26.4−2.2 3 1732 76 7 15.5−1.9 +3.0 ± +2.4 NGC5548 23.4−2.3 3 1980 30 7 17.9−1.8 +1.3 ± +1.0 NGC5548 8.7−3.8 3 1969 48 7 6.5−2.9 +0.5 ± +0.8 NGC5548 8.6−0.5 3 2173 89 7 7.9−0.8 +1.0 ± +4.2 NGC5548 16.0−1.2 3 3078 197 7 29.6−4.4 +0.6 ± +3.5 NGC5548 5.5−0.7 4 4270 292 8,9 19.6−3.6 +0.1 ± +0.5 NGC6814 7.4−0.1 4 1610 108 8,9 3.7−0.5 +0.5 ± +1.4 NGC7469 11.5−0.7 3 1456 207 7 4.8−1.4 +2.2 ± +2.6 PG1229+204 40.3−1.0 3 1385 111 7 15.1−2.4 +12.0 ± +8.0 PG1411+442 49.1−4.9 3 1607 169 7 24.7−5.8 +6.4 ± +63.2 PG1426+015 148.7−10.2 3 3442 308 7 343.6−65.9 +27.9 ± +41.2 PG1617+175 79.3−5.3 3 2626 211 7 106.6−18.6 +4.0 ± +3.0 PG2130+099 31.0−4.0 6 1825 65 1 20.1−3.0 +0.9 ± +0.5 SBS1116+583A 2.4− 4 1528 184 8,9 1.1− 0.9 ± 0.5
Note. — See Table 6.5 for details.
Table 6.7. Reverberation Measurements and Updated Virial Products for the Third Part of the AGN Sample
156 b c Galaxy Mvir MBH σ∗ σ∗ a 6 6 −1 Classification (10 M⊙) (10 M⊙) (km s ) Reference
+1.2 +5.1 3C 120 Classical 12.2−1.2 51.1−5.1 162 20 1 +36 +150.8 ± 3C 390.3 Classical 260−23 1089−96.4 273 16 2 +3.9 +16.2 ± Ark 120 Classical 22.7−5.5 95.1−23.1 221 17 2 +0.2 +1.4 ± Arp 151 Classical 1.1−0.1 4.6−1.2 118 4 3 +0.9 +3.8 ± Mrk 50 Classical 6.2−0.9 25.7−3.9 109 14 4 +4.4 +18.3 ± Mrk 79 Barred Pseudobulge 18.7−7.3 78.5−30.4 130 12 2 +1.3 +5.5 ± Mrk 110 Pseudobulge 5.0−2.0 20.9−8.6 91 7 5 +0.1 +1.1 ± Mrk 202 Classical 0.3−0.1 1.2−1.1 78 3 3 +1.1 +4.5 ± Mrk 279 Pseudobulge 7.0−1.0 29.3−4.5 197 12 2 +1.0 +4.1 ± Mrk 509 Classical 21.5−1.0 89.9−4.1 184 12 6 +1.2 +5.0 ± Mrk 590 Pseudobulge 7.1−1.6 29.8−6.7 189 6 2 +2.2 +9.1 ± Mrk 817 Barred Pseudobulge 14.1−2.4 59.3−10.2 120 15 2 +0.2 +1.4 ± Mrk 1310 Pseudobulge 0.5−0.2 2.0−1.3 84 5 3 +2.0 +8.4 ± NGC 3227 Barred Pseudobulge 5.2−2.1 21.6−8.7 136 4 2 +0.7 +3.1 ± NGC 3516 Barred Pseudobulge 7.1−0.6 29.9−2.5 181 5 2 +0.7 +3.2 ± NGC 3783 Barred Pseudobulge 4.3−0.8 18.2−3.6 95 10 7 +0.1 +1.1 ± NGC 4051 Barred Pseudobulge 0.5−0.1 2.2−1.1 89 3 2 +0.9 +4.0 ± NGC 4151 Barred Pseudobulge 8.4−0.5 35.1−2.3 97 3 2 +0.2 +1.5 ± NGC 4253 Barred Pseudobulge 0.3−0.2 1.2−1.5 93 32 3 +0.4 +1.9 ± NGC 4593 Barred Pseudobulge 2.1−0.3 8.9−1.7 135 6 2 +0.2 +1.4 ± NGC 4748 Barred Pseudobulge 0.7− 3.1− 105 13 3 0.2 1.4 ±
(cont’d)
Table 6.8. The AGN MBH–σ∗ Sample
157 Table 6.8—Continued
b c Galaxy Mvir MBH σ∗ σ∗ a 6 6 −1 Classification (10 M⊙) (10 M⊙) (km s ) Reference
+1.7 +7.9 NGC 5548 Pseudobulge 13.6−2.0 63.4−9.1 195 13 3 +0.5 +2.4 ± NGC 6814 Barred Pseudobulge 3.7−0.5 15.6−2.4 95 3 3 +1.4 +5.8 ± NGC 7469 Barred Pseudobulge 4.8−1.4 19.9−5.9 131 5 2 +2.6 +10.7 ± PG 1229+204 Barred Pseudobulge 15.1−2.4 63.1−10.3 162 32 8 +8.0 +33.5 ± PG 1411+442 Classical 24.7−5.8 103.6−24.1 209 30 6 +63.2 +265 ± PG 1426+015 Classical 343.6−65.9 1440−276 217 15 9 +41.2 +172.7 ± PG 1617+175 Classical 106.6−18.6 446.7−77.8 201 37 6 +3.0 +12.5 ± PG 2130+099 Pseudobulge 20.1−3.0 84.2−12.5 163 19 6 +0.5 +2.3 ± SBS 1116+583A Barred Pseudobulge 1.1− 4.6− 92 4 3 0.5 2.3 ±
Note. — Classifications were made using the host galaxy decompositions of Bentz et al. (2009a) and the critera are discussed in the text.For objects with multiple reverberation measurements, the adopted virial product is the average of the logarithm of the different virial products.. All MBH were computed using f = 4.19. Velocity Dispersion References: 1. Nelson & Whittle 1995; 2. Nelson et al. 2004; 3. Woo et al. 2010; 4. Barth et al. 2011; 5. Ferrarese et al. 2001; 6. This work; 7. Onken et al. 2004; 8. Dasyra et al. 2007; 9. Watson et al. 2008.
158 Chapter 7
Conclusions
7.1. Summary
In this dissertation, I have discussed the results of a major observational effort to investigate the size and dynamics of the BLR and improve measurements of
BH masses in AGNs. Accurate BH masses in AGNs are extremely important, as
AGNs currently provide the only means for exploring the BH population and the connection between BHs and their host galaxies at large distances.
In Chapter 2, we discussed our results from a reverberation campaign carried out in 2008, which yielded a measurement of the time lag of the Hβ line in
PG 2130+099. Previous measurements of τcent overestimated the size of the BLR and therefore the mass of the BH, most likely due to the undersampling in the light curves. A re-analysis of the previous data suggests that the true BLR size and MBH are consistent with our new values. The most recent reverberation measurements for PG 2130+099 presented in this study (Chapter 5) remove the discrepancies previously found for this object based on the MBH–σ∗ and RBLR–L relationships.
159 These results highlight the importance of adequate time sampling and data quality in reverberation mapping experiments, showing that the accuracy as well as precision suffers with the use of inadequate light curves.
In Chapter 3, we detailed our massive, 140-day reverberation mapping campaign carried out in the fall of 2010, which has yielded improvements in the reverberation mapping database. We successfully measured the average size of the Hβ-emitting region in five objects observed in the campaign. Four of these measurements constitute significant improvements in precision compared to previous measurements, and the fifth was the first reverberation measurement for the object.
We also measured the line widths in these objects and combined them with the measured lags to determine MBH for the sample. In all cases, our new measurements are consistent with previous measurements, but with reduced uncertainties. We placed our objects on the most current RBLR–L relationship and find that our new measurements place our objects in locations consistent with previous measurements when taking into account the poorer precision of past measurements and observed mean luminosity changes. This is consistent with the location of the BLR being regulated by photoionization physics. This work, like our previous work with
PG 2130+099, demonstrates the importance of adequately-sampled light curves in reducing uncertainties in BLR radius measurements.
In addition to our measurements of the Hβ-emitting region, we have also presented the first robust He ii λ4686 reverberation lag measurement in a NLS1
160 galaxy, Mrk 335. As described in Chapter 4, the MBH measurements from He ii and
Hβ are consistent with one another, suggesting that the gas producing the He ii emission resides in a similar structure as that producing Hβ emission. While other high-ionization lines such as C iv show evidence for outflows, we do not see this in He ii, possibly because the He ii-emitting gas does not arise cospatially with gas producing C iv emission. This has practical implications for future reverberation efforts, as the He ii emission, with its shorter time lags, may allow us to more efficiently measure MBH in objects at high redshift.
Using this reverberation data set, we have for the first time obtained velocity- delay maps constructed from the line and continuum variations observed in four objects from our 2010 reverberation campaign. I discussed this work in detail in
Chapter 5. These maps provide new insights into the structure of the BLR and constitute a dramatic increase in the number of objects that have at least some information on the velocity field of the BLR. Along with the velocity-delay maps for Arp 151 (Bentz et al. 2010b) and the models of Mrk 50 (Pancoast et al. 2012), these velocity-delay maps provide the strongest constraints on the structure of the
BLR. Our velocity-delay map for 3C 120 shows very similar structure to the map of
Arp 151, which also shows a lack of prompt response in the Balmer lines. We also see very asymmetric profiles in both the Balmer and high-ionization emission that is suggestive of infalling gas in all of our objects. In 3C 120 and Mrk 335, the transfer function structure for the Balmer lines differs from that for the He ii λ4686 emission
161 line, suggesting different structures dominating at different BLR radii. In all cases where our data are of sufficient quality to constrain the structure of the BLR, we see clear evidence of infall and rotation, both of which result from the gravitational influence of the BH. As gravitationally dominated motion is the key assumption of reverberation mapping, our results strongly support the reliability of MBH estimates derived from reverberation mapping.
Measurements of stellar velocity dispersions in galaxies provide an important means of examining the coevolution of BHs and their host galaxies. In Chapter 6 of this work, we presented new stellar velocity dispersion measurements for four luminous quasars (PG1411+442, PG1617+175, Mrk 509, and PG2130+099) with the NIFS instrument and the ALTAIR laser guide star adaptive optics system on the
Gemini North 8-m telescope for the purpose of examining the MBH–σ∗ relationship at the high-mass end of the distribution. The four measurements are consistent with the MBH–σ∗ relation exhibited by lower-luminosity AGNs with lower-mass black holes. We find no offset at higher mass, nor do we see correlations with galaxy morphology. As part of this analysis, we have recalculated the virial products for the entire sample of reverberation-mapped AGNs and used these data to redetermine the mean virial factor f that places the reverberation data on the quiescent M –σ∗ h i BH relation. With our updated measurements and new additions to the AGN sample, we obtain f = 4.19 1.08, which is slightly lower than, but consistent with, most h i ± previous determinations.
162 7.2. Future Work
Together, the results presented in this dissertation constitute a substantial improvement in the precision and accuracy of reverberation-based MBH measurements as well as our understanding of the structure of the BLR. The Hβ RBLR–L relation is now characterized quite well. However, there is still much to do. Below I highlight some useful potential directions for future explorations of AGN dynamics and masses.
7.2.1. Modeling the Structure of the BLR
There are now five objects for which relatively detailed velocity-delay maps have been recovered and from these we have gained much insight into the structure of the BLR. In all five objects, we see initial signs of infalling and virial components to the BLR, which directly supports the use of the BLR to measure MBH in AGNs.
However, we have not yet fit models to these data to determine the detailed structure of the gas. While continuing to observe more targets to add to the sample of high-quality velocity-delay maps, the next major step is to model these data directly to determine the geometry and possibly the orientation of the BLR components themselves, rather than just searching for qualitative signatures in the velocity-delay maps. This type of modeling has been done for one AGN by Pancoast et al. (2012), who were able to fit a simple BLR model to the light curves and spectra of Mrk 50.
163 However, the BLR appears to be quite complex and likely requires more detailed modeling to accurately characterize its structure, and future reverberation mapping efforts would benefit from increased focus on this problem.
7.2.2. The RBLR–L Relation and BLR structure in
High-Ionization Lines
While the Hβ RBLR–L relation is now very well-established, accurate RBLR–L relations are still required for other emission lines used in MBH calculations. For higher-redshift objects in which the Hβ emission line is redshifted out of the optical, we must turn to the Mg iiλ2798 and C ivλ1549 emission lines to estimate MBH.
At redshifts between about 1 and 5, C ivλ1549 can be observed from the ground and is suitable for use because of its strength and minimal blending with nearby features. However, using C iv (or Mg ii) to estimate MBH with confidence first requires reverberation mapping studies to establish its RBLR–L relation and show that it behaves well enough to use for MBH measurements. To date, very few studies have been able to successfully measure RBLR using the C iv emission line, and there is just one reliable measurement for Mg ii (Metzroth et al. 2006). Kaspi et al. (2007) provide the most current RBLR–L relation for C iv, which is based on very few measurements, many of which are dubious. The gaps in the relation must be filled with high-quality reverberation mapping data for the C iv emission
164 line to reliably establish a C iv RBLR–L relation. The use of single-epoch virial masses is currently the only way to estimate MBH masses in AGNs at high redshift, so this information is essential to any future efforts to obtain MBH estimates in distant AGNs. The accuracy of MBH estimates in high-redshift objects would benefit greatly from a long-duration ground-based reverberation campaign aimed at the C iv emission line in high-redshift objects (z =1.5 3.5). From this we can populate the − high-luminosity end of the C iv RBLR–L relation as well as gain information about the structure of the C iv-emitting region of the BLR.
In addition, there has been much recent discussion in the literature regarding the suitability of C iv for single-epoch MBH measurements in general (e.g., Assef et al. 2011; Trakhtenbrot & Netzer 2012; Denney 2012). Some have suggested that the motion of the C iv-emitting gas in the BLR is not necessarily dominated by gravity (e.g., Richards et al. 2002). C iv has also been shown to have a significant non-varying component that can affect MBH estimates made from a single spectrum
(Denney 2012). However, with data of sufficient quality, we can correct for these various factors and obtain MBH measurements consistent with those made from Hβ
(e.g., Assef et al. 2011). Whether C iv is inherently unusable for MBH measurements, or whether it just requires more care when using it is still under debate, as many of the studies condemning the use of C iv altogether were completed with data that were quite poor. High-quality reverberation data sets will allow us to investigate the use of C iv in MBH measurements, and in particular, investigate the observed
165 non-varying component of the C iv line found by Denney (2012) by looking at the RMS residual spectra to assess the effect of the non-variable component on single-epoch MBH measurements. Future efforts will benefit from space-based UV and simultaneous ground-based optical reverberation campaigns targeted at the higher-ionization emission lines to recover velocity-delay maps.
7.2.3. The MBH–σ Relation and the AGN MBH Scale ∗
Finally, as I note in Chapter 6, there is still a lack of secure σ∗ measurements in high-luminosity quasars to fully populate the MBH–σ∗ relation. We appear to have reached the limit of our current equipment due to a combination of telluric and sky contamination issues as well as quasar contamination. Future efforts to populate the high-luminosity end of the relation are still needed, and will likely be successful with improvements in AO systems as well as the use of space-based equipment such as the upcoming JWST . Eliminating the telluric absorption and sky line contamination alone will drastically improve spectra in the NIR. As this contamination was prohibitive in our attempts with several objects, JWST has the potential to help greatly in our attempts to measure σ∗ in these high-redshift objects.
While we do not see any current evidence for deviations from the quiescent
MBH–σ∗ relation in AGNs, both theoretical and observational studies are still
166 reporting possible morphological and environmental dependencies in the quiescent relation (e.g., Graham & Scott 2013; King 2010; Zubovas & King 2012). This leads us to wonder whether or not the MBH–σ∗ relation is truly the most suitable relation to use for the calibration of the AGN MBH scale. The MBH–LBulge relation for AGNs has recently been shown to have very little scatter (Bentz et al. 2009b), and the relation for quiescent galaxies is expected to have similarly small scatter.
As such, the quiescent MBH–LBulge relation may be more well-suited to do so than the MBH–σ∗ relation. There have been recent efforts to refine the MBH–LBulge relation in quiescent galaxies using MBH and LBulge measurements compiled from the literature and using SDSS photometry (e.g., Beifiori et al. 2012), but we suspect that using a homogeneous data set with a uniform method to measure LBulge will reduce the scatter in this relation even more. Possible improvements in the AGN
MBH scale could be made by using a refined, homogenous set of LBulge measurements to establish an updated MBH–LBulge relation, which can then be used to obtain a new f factor. The updated f factor will provide a more accurate calibration h i h i for AGN MBH measurements, and the improved MBH–LBulge relation will also yield information about the true intrinsic scatter of the relation and provide further constraints and insight into the evolution of galaxies.
167 BIBLIOGRAPHY
Alard, C. 2000, A&AS, 144, 363
Alard, C., & Lupton, R. H. 1998, ApJ, 503, 325
Alexander, T. 1997, in Astronomical Time Series, ed. D. Maoz, A. Sternberg, & E.M. Leibowitz (Dordrecht: Kluwer), 163
Antokhin, I. I., & Bochkarev, N. G. 1983, AZh, 60, 448
Assef, R. J., et al. 2011, ApJ, 742, 93
Barth, A. J., et al. 2011, ApJ, 743, L4
Bahcall, J. N., Kozlovsky, B.-Z., & Salpeter, E. E. 1972, ApJ, 171, 467
Beifiori, A., Courteau, S., Corsini, E. M., & Zhu, Y. 2012, MNRAS, 419, 2497
Bentz, M.C., Peterson, B.M., Pogge, R.W., Vestergaard, M., & Onken, C.A. 2006a, ApJ, 644, 133
Bentz, M. C. et al. 2006b, ApJ, 651, 775
Bentz, M. C., Peterson, B. M., Netzer, H., Pogge, R. W., & Vestergaard, M. 2009a, ApJ, 697, 160
Bentz, M. C., Peterson, B. M., Pogge, R. W., & Vestergaard, M. 2009b, ApJ, 694, L166
Bentz, M. C., et al. 2009c, ApJ, 705, 199
Bentz, M. C., et al. 2010a, ApJ, 716, 993
Bentz, M. C., et al. 2010b, ApJ, 720, L46
Bentz, M. C., et al. 2013, ApJ, 767, 149
Bertin, E., & Arnouts, S. 1996, A&AS, 117, 393
168 Blandford, R. D., & McKee, C. F. 1982, ApJ, 255, 419
Bochkarev, N. G., & Antokhin, I. I. 1982, Astronomicheskij Tsirkulyar, 1238, 1 Boller, T., Brandt, W. N., & Fink, H. 1996, A&A, 305, 53
Boroson, T. A. 2011, in Narrow-Line Seyfert 1 Galaxies and their Place in the Universe
Boyle, B. J., Georgantopoulos, I., Blair, A. J., Stewart, G. C., Griffiths, R. E., Shanks, T., Gunn, K. F., & Almaini, O. 1998, MNRAS, 296, 1
Brewer, B. J., et al. 2011, ApJ, 733, L33
Cappellari, M., & Emsellem, E. 2004, PASP, 116, 138
Capriotti, E. R., Foltz, C. B., & Peterson, B. M. 1982, ApJ, 261, 35
Christou, J. C., Neichel, B., Rigaut, F., Sheehan, M., McDermid, R. M., & Trancho, G. 2010, in Adaptive Optics Systems II, SPIE Proc, 7736, 77361R
Clavel, J., Wamsteker, W., & Glass, I. S. 1989, ApJ, 337, 236
Clavel, J., et al. 1990, MNRAS, 246, 668
Clavel, J., et al. 1991, ApJ, 366, 64
Clavel, J. 1991, Variability of Active Galaxies, ed. C. M. Duschl, S. J. Wagner, & M. Camenzind (Berlin: Springer-Verlag), 31
Crenshaw, D. M., & Blackwell, Jr., J. H. 1990, ApJ, 358, L37
Collin, S., Kawaguchi, T., Peterson, B.M., & Vestergaard, M. 2006, A&A, 456, 75
Dasyra, K. M., et al. 2007, ApJ, 657, 102
Davidson, K. 1972, ApJ, 171, 213
Davidson, K., & Netzer, H. 1979, Reviews of Modern Physics, 51, 715
Denney, K. D., et al. 2006, ApJ, 653, 152
Denney, K. D., et al. 2009a, ApJ, 702, 1353
Denney, K. D., et al. 2009b, ApJ, 704, L80
Denney, K. D., et al. 2010, ApJ, 721, 715
Denney, K., et al. 2011, in Narrow-Line Seyfert 1 Galaxies and their Place in the Universe, Proceedings of Science, POS(NLS1)032
169 Denney, K. D. 2012, ApJ, 759, 44
Dietrich, M., et al. 2012, ApJ, 757, 53
Dietrich, M., et al. 1993, ApJ, 408, 416
Di Matteo, T., Colberg, J., Springel, V., Hernquist, L., & Sijacki, D. 2008, ApJ, 676, 33
Di Matteo, T., Springel, V., & Hernquist, L. 2005, Nature, 433, 604
Doroshenko, V. T., & Sergeev, S. G. 2003, A&A, 405, 909
Doroshenko, V. T., Sergeev, S. G., Klimanov, S. A., Pronik, V. I., & Efimov, Y. S. 2012, MNRAS, 426, 416
Edelson, R.A., & Krolik, J.H. 1988, ApJ, 333, 646
Fabian, A. C. 1999, MNRAS, 308, L39
Ferland, G. J., Peterson, B. M., Horne, K., Welsh, W. F., & Nahar, S. N. 1992, ApJ, 387, 95
Ferrarese, L., & Merritt, D. 2000, ApJ, 539, L9
Ferrarese, L., Pogge, R.W., Peterson, B.M., Merritt, D., Wandel, A., & Joseph, C.L. 2001, ApJ, 555, L79
Frieman, J. A., et al. 2008, AJ, 135, 338
Franceschini, A., Hasinger, G., Miyaji, T., & Malquori, D. 1999, MNRAS, 310, L5
Fromerth, M.J., & Melia, F. 2000, ApJ, 533, 172
Gaskell, C. M., & Sparke, L. S. 1986, ApJ, 305, 175
Gaskell, C. M., & Peterson, B. M. 1987, ApJS, 65, 1
Gaskell, C. M. 1988, ApJ, 325, 114
Gebhardt, K., et al. 2000, ApJ, 539, L13
Gebhardt, K., et al. 2000b, ApJ, 543, L5
Goodrich, R. W. 1989, ApJ, 342, 224
Graham, A. W., Onken, C. A., Athanassoula, E., & Combes, F. 2011, MNRAS, 412, 2211
170 Graham, A. W., & Scott, N. 2013, ApJ, 764, 151
Greene, J. E., & Ho, L. C. 2007, ApJ, 667, 131
Greene, J. E., et al. 2010, ApJ, 721, 26
Grier, C. J., et al. 2008, ApJ, 688, 837
Grier, C. J., et al. 2012a, ApJ, 744, L4
Grier, C. J., et al. 2012b, ApJ, 755, 60
Grier, C. J., et al. 2013, ApJ, 764, 47
Guerras, E., Mediavilla, E., Jimenez-Vicente, J., Kochanek, C. S., Mu˜noz, J. A., Falco, E., & Motta, V. 2013, ApJ, 764, 160
G¨ultekin, K., et al. 2009, ApJ, 698, 198
Hopkins, P. F., Murray, N., & Thompson, T. A. 2009, MNRAS, 398, 303
Horne, K., Welsh, W. F., & Peterson, B. M. 1991, ApJ, 367, L5
Horne, K. 1994, in Astronomical Society of the Pacific Conference Series, Vol. 69, Reverberation Mapping of the Broad-Line Region in Active Galactic Nuclei, ed. P. M. Gondhalekar, K. Horne, & B. M. Peterson, 23–25
Horne, K., Peterson, B. M., Collier, S. J., & Netzer, H. 2004, PASP, 116, 465
Jahnke, K., & Macci`o, A. V. 2011, ApJ, 734, 92
Jorgensen, I., Franx, M., & Kjaergaard, P. 1995, MNRAS, 276, 1341
Kaspi, S., Smith, P. S., Netzer, H., Maoz, D., Jannuzi, B. T., & Giveon, U. 2000, ApJ, 533, 631
Kaspi, S., Maoz, D., Netzer, H., Peterson, B.M., Vestergaard, M., & Jannuzi, B.T. 2005, ApJ, 629, 61
Kaspi, S., Brandt, W. N., Maoz, D., Netzer, H., Schneider, D. P., & Shemmer, O. 2007, ApJ, 659, 997
Kassebaum, T. M., Peterson, B. M., Wanders, I., Pogge, R. W., Bertram, R., & Wagner, R. M. 1997, ApJ, 475, 106
Kauffmann, G., et al. 2003, MNRAS, 346, 1055
Kelly, B. C., Bechtold, J., & Siemiginowska, A. 2009, ApJ, 698, 895
171 Kelly, B. C., Vestergaard, M., Fan, X., Hopkins, P., Hernquist, L., & Siemiginowska, A. 2010, ApJ, 719, 1315
King, A. 2003, ApJ, 596, L27
King, A. 2005, ApJ, 635, L121
King, A. R. 2010, MNRAS, 402, 1516
Kollatschny, W. 2003, A&A, 407, 461
Kollatschny, W., & Dietrich, M. 1997, A&A, 323, 5
Kollmeier, J. A., et al. 2006, ApJ, 648, 128
Komossa, S. 2008, Rev. Mex. AA Ser. Conf., 32, 86
Koratkar, A. P., & Gaskell, C. M. 1989, ApJ, 345, 637
Kormendy, J., & Richstone, D. 1995, ARA&A, 33, 581
Kormendy, J., & Kennicutt, Jr., R. C. 2004, ARA&A, 42, 603
Koz lowski, S., et al. 2010, ApJ, 708, 927
Krolik, J. H., Horne, K., Kallman, T. R., Malkan, M. A., Edelson, R. A., & Kriss, G. A. 1991, ApJ, 371, 541
Leighly, K. M., & Moore, J. R. 2004, ApJ, 611, 107
Lauer, T. R., Tremaine, S., Richstone, D., & Faber, S. M. 2007, ApJ, 670, 249
MacLeod, C. L., et al. 2010, ApJ, 721, 1014
MacLeod, C. L., et al. 2012, ApJ, 753, 106
Magorrian, J., et al. 1998, AJ, 115, 2285
Marziani, P., & Sulentic, J. W. 2012, New A Rev., 56, 49
Mathur, S., Fields, D., Peterson, B. M., & Grupe, D. 2012, ApJ, 754, 146
McConnell, N. J., & Ma, C.-P. 2013, ApJ, 764, 184
McConnell, N. J., Ma, C.-P., Gebhardt, K., Wright, S. A., Murphy, J. D., Lauer, T. R., Graham, J. R., & Richstone, D. O. 2011, Nature, 480, 215
Merloni, A., Rudnick, G., & Di Matteo, T. 2004, MNRAS, 354, L37
Metzroth, K. G., Onken, C. A., and Peterson, B. M. 2006, ApJ, 647, 901
172 McGill, K.L., Woo, J.-H., Treu, T., Malkan, M.A. 2008, ApJ, 673, 703
Morgan, C. W., Kochanek, C. S., Morgan, N. D., & Falco, E. E. 2010, ApJ, 712, 1129
Murray, N., Quataert, E., & Thompson, T. A. 2005, ApJ, 618, 569
Nelson, C.H., Green, R.F., Bower, G., Gebhardt, K., & Weistrop, D. 2004, ApJ, 615, 652
Nelson, C. H., & Whittle, M. 1995, ApJS, 99, 67
Onken, C. A., Ferrarese, L., Merritt, D., Peterson, B. M., Pogge, R. W., Vestergaard, M., & Wandel, A. 2004, ApJ, 615, 645
Osterbrock, D. E., & Pogge, R. W. 1985, ApJ, 297, 166
Pancoast, A., Brewer, B. J., & Treu, T. 2011, ApJ, 730, 139
Pancoast, A., et al. 2012, ApJ, 754, 49
Park, D., Kelly, B. C., Woo, J.-H., & Treu, T. 2012, ApJS, 203, 6
Peng, C. Y. 2007, ApJ, 671, 1098
Peng, C. Y. 2010, arXiv:1002.2664
Perez, E., Robinson, A., & de La Fuente, L. 1992, MNRAS, 255, 502
Peterson, B. M., et al. 1991, ApJ, 368, 119
Peterson, B. M. 1993, PASP, 105, 247
Peterson, B. M., Pogge, R. W., Wanders, I., Smith, S. M., & Romanishin, W. 1995, PASP, 107, 579
Peterson, B. M., Wanders, I., Bertram, R., Hunley, J. F., Pogge, R. W., & Wagner, R. M. 1998a, ApJ, 501, 82
Peterson, B.M., Wanders, I., Horne, K., Collier, S., Alexander, T., & Kaspi, S. 1998b, PASP, 110, 660
Peterson, B. M., et al. 2000, ApJ, 542, 161
Peterson, B. M., et al. 2004, ApJ, 613, 682
Peterson, B. M. 2011, in Narrow-Line Seyfert 1 Galaxies and their Place in the Universe, Proceedings of Science, POS(NLS1)034
173 Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery, B. P. 1992, Numerical recipes in FORTRAN. The art of scientific computing (Cambridge: University Press, 2nd ed.), 660
Richards, G. T., Vanden Berk, D. E., Reichard, T. A., Hall, P. B., Schneider, D. P., SubbaRao, M., Thakar, A. R., & York, D. G. 2002, AJ, 124, 1 Richards, G. T., et al. 2011, AJ, 141, 167
Rodriguez-Pascual, P. M., et al. 1997, ApJS, 110, 9
Sanders, D. B., Soifer, B. T., Elias, J. H., Madore, B. F., Matthews, K., Neugebauer, G., & Scoville, N. Z. 1988, ApJ, 325, 74
Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525
Sergeev, S. G., Pronik, V. I., Sergeeva, E. A., & Malkov, Y. F. 1999, ApJS, 121, 159
Sergeev, S. G., Doroshenko, V. T., Golubinskiy, Y. V., Merkulova, N. I., & Sergeeva, E. A. 2005, ApJ, 622, 129
Sergeev, S. G., Doroshenko, V. T., Dzyuba, S. A., Peterson, B. M., Pogge, R. W., & Pronik, V. I. 2007, ApJ, 668, 708
Shappee, B. J., & Stanek, K. Z. 2011, ApJ, 733, 124
Shen, Y., Greene, J. E., Strauss, M. A., Richards, G. T., & Schneider, D. P. 2008, ApJ, 680, 169
Silk, J., & Rees, M. J. 1998, A&A, 331, L1
Silverman, J. D., et al. 2008, ApJ, 675, 1025
Sulentic, J. W., Zwitter, T., Marziani, P., & Dultzin-Hacyan, D. 2000, ApJ, 536, L5
Th´evenaz, P., Blu, T., & Unser, M. 2000, IEEE Transactions on Medical Imaging, 19, 739
Tody, D. 1986, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 627, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, ed. D. L. Crawford, 733
Trakhtenbrot, B., & Netzer, H. 2012, MNRAS, 427, 3081 Tremaine, S., et al. 2002, ApJ, 574, 740
Ulrich, M.-H., & Horne, K. 1996, MNRAS, 283, 748
Vacca, W. D., Cushing, M. C., & Rayner, J. T. 2003, PASP, 115, 389
174 van Groningen, E., & Wanders, I. 1992, PASP, 104, 700
Veilleux, S., et al. 2009, ApJS, 182, 628
Vestergaard, M., Fan, X., Tremonti, C. A., Osmer, P. S., & Richards, G. T. 2008, ApJ, 674, L1
Vestergaard, M., & Osmer, P. S. 2009, ApJ, 699, 800
Wanders, I., et al. 1995, ApJ, 453, L87
Wanders, I., & Peterson, B. M. 1996, ApJ, 466, 174
Watson, D., Denney, K. D., Vestergaard, M., & Davis, T. M. 2011, ApJ, 740, L49
Watson, L.C. et al. 2008, ApJL, 682, 21
Welsh, W. F. 1999, PASP, 111, 1347
White, R. J., & Peterson, B. M. 1994, PASP, 106, 879
Whittle, M. 1992, ApJS, 79, 49
Woo, J.-H., et al. 2010, ApJ, 716, 269
Yu, Q., & Tremaine, S. 2002, MNRAS, 335, 965
Zu, Y., Kochanek, C. S., & Peterson, B. M. 2011, ApJ, 735, 80
Zu, Y., Kochanek, C. S., Koz lowski, S., & Udalski, A. 2013, ApJ, 765, 106
Zubovas, K., & King, A. R. 2012, MNRAS, 426, 2751
175 Appendix A
2010 Reverberation Campaign Light Curves
176 Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5431.45 A 6.01 0.07 5430.48 A 1.78 0.02 5430.58 A 3.05 0.09 5430.54 A 7.93 0.07 5430.38 A 3.56 0.05 ± ± ± ± ± 5436.45 A 5.98 0.09 5431.48 A 1.78 0.03 5432.57 A 2.97 0.05 5431.52 A 7.83 0.09 5431.38 A 3.67 0.06 ± ± ± ± ± 5438.46 A 6.04 0.08 5432.48 A 1.77 0.03 5433.55 A 2.91 0.06 5432.51 A 7.85 0.11 5432.38 A 3.55 0.07 ± ± ± ± ± 5440.90 M 6.47 0.09 5433.41 W 1.85 0.02 5436.56 A 2.92 0.05 5433.49 A 7.80 0.13 5433.32 A 3.56 0.07 ± ± ± ± ± 5441.86 M 6.19 0.08 5433.43 A 1.78 0.04 5436.60 W 3.09 0.01 5435.57 W 7.65 0.02 5433.36 W 3.38 0.03 ± ± ± ± ± 5442.83 M 6.14 0.08 5437.47 A 1.73 0.03 5437.55 A 2.93 0.04 5436.48 A 7.61 0.10 5434.37 A 3.52 0.06 ± ± ± ± ± 5444.84 M 6.24 0.08 5438.42 A 1.79 0.05 5437.57 W 3.00 0.02 5436.58 W 7.53 0.02 5435.39 A 3.51 0.04 177 ± ± ± ± ± 5445.85 M 6.24 0.08 5439.46 A 1.77 0.05 5439.58 A 2.87 0.08 5437.48 A 7.50 0.09 5435.41 W 3.48 0.01 ± ± ± ± ± 5447.92 M 6.24 0.08 5440.45 A 1.77 0.03 5441.95 M 2.90 0.09 5437.57 W 7.47 0.02 5436.32 W 3.47 0.01 ± ± ± ± ± 5449.97 M 6.21 0.08 5440.46 W 1.77 0.01 5443.94 M 3.07 0.10 5440.57 W 7.48 0.02 5436.36 A 3.49 0.05 ± ± ± ± ± 5450.47 A 5.98 0.04 5440.95 M 1.73 0.04 5444.52 A 3.04 0.04 5442.00 M 7.22 0.05 5437.33 W 3.46 0.01 ± ± ± ± ± 5451.76 M 6.06 0.08 5441.91 M 1.75 0.04 5445.48 A 2.98 0.03 5442.95 M 7.19 0.05 5437.37 A 3.50 0.04 ± ± ± ± ± 5452.47 A 5.77 0.11 5442.29 W 1.84 0.01 5446.98 M 3.02 0.10 5443.41 A 7.18 0.07 5438.30 A 3.44 0.04 ± ± ± ± ± 5453.83 M 6.02 0.08 5442.90 M 1.77 0.04 5447.50 W 3.14 0.01 5443.53 W 7.02 0.02 5438.38 W 3.46 0.01 ± ± ± ± ± 5454.45 A 5.84 0.05 5443.36 W 1.76 0.01 5447.53 A 3.00 0.03 5443.55 C 7.08 0.12 5439.27 W 3.42 0.01 ± ± ± ± ±
(cont’d) Table A.1. V -band and Continuum Fluxes Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5454.78 M 5.93 0.08 5443.38 A 1.79 0.02 5447.96 M 3.01 0.09 5444.40 A 7.12 0.07 5439.41 A 3.37 0.05 ± ± ± ± ± 5455.44 A 5.84 0.05 5443.51 C 1.72 0.03 5448.56 A 2.97 0.03 5444.96 M 7.07 0.05 5440.35 W 3.44 0.01 ± ± ± ± ± 5456.46 A 5.88 0.05 5444.31 W 1.75 0.01 5450.58 A 3.12 0.03 5446.52 W 7.15 0.01 5440.36 A 3.38 0.03 ± ± ± ± ± 5456.89 M 5.89 0.08 5444.38 A 1.77 0.02 5451.58 A 3.09 0.03 5447.55 W 7.00 0.01 5441.40 W 3.44 0.01 ± ± ± ± ± 5457.77 M 5.95 0.08 5444.46 C 1.70 0.03 5452.95 M 3.03 0.10 5447.57 A 6.97 0.07 5441.73 M 3.45 0.03 ± ± ± ± ±
178 5458.84 M 5.99 0.08 5444.89 M 1.72 0.04 5454.54 A 3.12 0.03 5449.57 A 7.16 0.10 5442.27 W 3.46 0.01 ± ± ± ± ± 5459.44 A 5.79 0.08 5445.39 A 1.76 0.02 5454.90 M 3.02 0.10 5450.50 C 6.98 0.12 5442.71 M 3.44 0.03 ± ± ± ± ± 5463.36 A 5.86 0.12 5446.52 W 1.71 0.01 5455.58 A 3.11 0.03 5450.54 A 7.07 0.06 5443.32 A 3.43 0.03 ± ± ± ± ± 5466.85 M 6.10 0.08 5446.93 M 1.74 0.04 5455.88 M 3.06 0.10 5450.96 M 6.99 0.05 5443.34 W 3.43 0.01 ± ± ± ± ± 5467.89 M 5.99 0.08 5447.33 W 1.75 0.01 5456.51 C 3.13 0.10 5451.49 C 7.16 0.12 5443.43 C 3.44 0.07 ± ± ± ± ± 5468.84 M 5.95 0.08 5447.49 A 1.78 0.02 5456.57 A 3.12 0.04 5453.98 M 7.16 0.05 5443.76 M 3.45 0.03 ± ± ± ± ± 5469.82 M 6.10 0.08 5448.33 W 1.73 0.01 5457.49 C 3.18 0.10 5454.50 A 7.08 0.08 5444.27 W 3.44 0.01 ± ± ± ± ± 5470.91 M 6.28 0.08 5448.46 A 1.75 0.02 5457.52 A 3.12 0.04 5454.97 M 7.17 0.05 5444.32 A 3.47 0.03 ± ± ± ± ± 5472.90 M 6.27 0.08 5449.33 W 1.72 0.01 5457.90 M 3.07 0.10 5455.55 A 7.18 0.08 5444.38 C 3.35 0.07 ± ± ± ± ± 5473.84 M 6.23 0.08 5449.54 A 1.75 0.02 5458.50 C 3.28 0.10 5455.97 M 7.25 0.05 5444.71 M 3.45 0.03 ± ± ± ± ±
(cont’d) Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5476.79 M 6.44 0.09 5449.90 M 1.80 0.05 5458.53 A 3.14 0.04 5456.40 C 7.01 0.12 5445.26 A 3.41 0.03 ± ± ± ± ± 5477.77 M 6.38 0.09 5450.29 W 1.71 0.01 5458.95 M 3.08 0.10 5456.54 A 7.40 0.08 5445.73 M 3.45 0.03 ± ± ± ± ± 5478.90 M 6.45 0.09 5450.44 A 1.73 0.02 5459.52 C 3.14 0.10 5456.96 M 7.31 0.05 5446.41 W 3.45 0.01 ± ± ± ± ± 5479.75 M 6.41 0.09 5450.90 M 1.79 0.04 5459.53 A 3.19 0.03 5457.41 C 7.16 0.12 5446.73 M 3.48 0.03 ± ± ± ± ± 5480.82 M 6.48 0.09 5451.45 C 1.67 0.03 5460.50 W 3.32 0.01 5457.45 A 7.42 0.08 5447.36 A 3.45 0.03 ± ± ± ± ±
179 5481.80 M 6.53 0.09 5451.50 A 1.75 0.02 5460.56 A 3.25 0.04 5457.96 M 7.37 0.05 5447.49 W 3.44 0.01 ± ± ± ± ± 5482.78 M 6.55 0.09 5451.84 M 1.71 0.04 5461.52 W 3.43 0.02 5458.50 A 7.48 0.08 5447.65 M 3.44 0.03 ± ± ± ± ± 5483.77 M 6.48 0.09 5452.44 A 1.71 0.02 5462.53 A 3.31 0.05 5458.54 C 7.47 0.13 5448.31 W 3.43 0.01 ± ± ± ± ± 5486.77 M 6.64 0.09 5452.44 C 1.68 0.03 5463.44 W 3.43 0.01 5459.48 C 7.43 0.13 5448.34 A 3.45 0.03 ± ± ± ± ± 5488.78 M 6.65 0.09 5453.53 A 1.70 0.03 5463.53 A 3.32 0.06 5459.50 A 7.55 0.09 5449.28 W 3.40 0.01 ± ± ± ± ± 5497.75 M 6.92 0.09 5453.89 M 1.73 0.04 5464.52 A 3.42 0.05 5460.51 A 7.69 0.12 5449.34 A 3.42 0.03 ± ± ± ± ± 5498.76 M 6.82 0.09 5454.31 W 1.67 0.01 5465.46 W 3.20 0.03 5460.51 W 7.53 0.02 5449.71 M 3.37 0.03 ± ± ± ± ± 5499.75 M 6.77 0.09 5454.39 C 1.69 0.03 5465.57 A 3.31 0.07 5461.49 W 7.44 0.02 5450.27 W 3.35 0.01 ± ± ± ± ± 5500.76 M 6.92 0.09 5454.43 A 1.70 0.02 5466.89 M 3.48 0.11 5462.59 A 7.72 0.08 5450.33 A 3.35 0.02 ± ± ± ± ± 5501.75 M 7.22 0.10 5454.84 M 1.66 0.04 5468.58 A 3.45 0.10 5463.46 W 7.37 0.02 5450.71 M 3.32 0.03 ± ± ± ± ±
(cont’d) Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5502.78 M 7.27 0.10 5455.41 A 1.67 0.01 5470.54 W 3.73 0.01 5463.49 A 7.74 0.11 5451.36 C 3.34 0.07 ± ± ± ± ± 5503.76 M 7.44 0.10 5455.83 M 1.65 0.04 5470.97 M 3.91 0.12 5463.94 M 7.84 0.06 5451.39 A 3.35 0.03 ± ± ± ± ± 5504.37 A 7.68 0.06 5456.37 C 1.63 0.03 5471.45 W 3.87 0.01 5464.44 A 7.87 0.10 5451.64 M 3.35 0.03 ± ± ± ± ± 5504.79 M 7.62 0.10 5456.43 A 1.69 0.02 5471.92 M 3.77 0.12 5465.45 W 7.68 0.03 5452.29 W 3.33 0.01 ± ± ± ± ± 5505.76 M 7.71 0.10 5457.38 C 1.66 0.03 5472.96 M 4.01 0.13 5465.60 A 7.80 0.14 5452.34 C 3.26 0.07 ± ± ± ± ±
180 5506.76 M 7.74 0.10 5457.44 A 1.69 0.02 5473.55 W 3.91 0.01 5466.96 M 7.84 0.06 5452.35 A 3.33 0.02 ± ± ± ± ± 5507.77 M 7.89 0.11 5457.84 M 1.63 0.04 5476.91 M 3.95 0.13 5467.48 W 7.76 0.02 5453.36 A 3.27 0.03 ± ± ± ± ± 5508.76 M 7.94 0.11 5458.30 W 1.66 0.01 5477.90 M 3.78 0.12 5467.97 M 7.92 0.06 5453.70 M 3.32 0.03 ± ± ± ± ± 5509.40 C 8.22 0.09 5458.47 A 1.64 0.02 5479.51 W 4.02 0.01 5468.42 C 7.72 0.13 5454.26 W 3.30 0.01 ± ± ± ± ± 5510.73 M 8.03 0.11 5458.47 C 1.61 0.03 5479.89 M 3.96 0.13 5468.53 W 7.91 0.02 5454.32 A 3.28 0.03 ± ± ± ± ± 5511.27 W 7.98 0.02 5458.90 M 1.66 0.04 5480.88 M 3.82 0.12 5468.98 M 7.78 0.06 5454.34 C 3.25 0.07 ± ± ± ± ± 5513.30 W 8.75 0.02 5459.29 W 1.63 0.01 5481.92 M 4.24 0.13 5469.49 W 7.87 0.02 5454.64 M 3.31 0.03 ± ± ± ± ± 5513.71 M 8.73 0.12 5459.41 A 1.71 0.04 5482.52 W 4.01 0.01 5470.52 W 7.82 0.02 5455.32 A 3.31 0.03 ± ± ± ± ± 5514.28 W 8.79 0.02 5460.34 W 1.63 0.01 5482.54 C 3.92 0.12 5471.49 W 7.78 0.01 5456.28 C 3.25 0.07 ± ± ± ± ± 5514.71 M 8.82 0.12 5462.36 A 1.48 0.05 5482.56 A 4.11 0.04 5471.98 M 8.03 0.06 5456.33 A 3.31 0.03 ± ± ± ± ±
(cont’d) Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5515.36 W 8.97 0.03 5465.45 A 1.61 0.04 5482.91 M 3.96 0.13 5473.56 W 8.01 0.01 5456.69 M 3.28 0.03 ± ± ± ± ± 5515.36 C 8.95 0.10 5466.27 W 1.52 0.02 5483.52 W 4.19 0.01 5474.45 W 8.23 0.01 5457.27 A 3.29 0.03 ± ± ± ± ± 5515.72 M 9.04 0.12 5466.44 A 1.60 0.04 5483.52 C 4.12 0.13 5476.97 M 8.19 0.06 5457.29 C 3.22 0.07 ± ± ± ± ± 5516.25 A 8.73 0.08 5467.27 W 1.54 0.01 5483.59 A 4.15 0.04 5477.97 M 8.14 0.06 5457.64 M 3.29 0.03 ± ± ± ± ± 5516.72 M 8.93 0.12 5468.38 C 1.51 0.03 5483.88 M 4.06 0.13 5478.96 M 8.30 0.06 5458.30 W 3.34 0.06 ± ± ± ± ±
181 5517.31 A 8.94 0.11 5468.44 A 1.57 0.02 5484.47 A 4.12 0.12 5479.48 W 8.39 0.01 5458.30 A 3.19 0.01 ± ± ± ± ± 5517.38 C 8.94 0.10 5468.49 W 1.54 0.01 5484.49 W 4.01 0.01 5479.96 M 8.30 0.06 5459.32 A 3.32 0.09 ± ± ± ± ± 5517.71 M 8.93 0.12 5468.87 M 1.61 0.04 5485.49 A 4.19 0.05 5480.94 M 8.46 0.06 5460.35 A 3.43 0.11 ± ± ± ± ± 5518.28 A 8.70 0.13 5469.31 W 1.52 0.01 5485.51 W 4.00 0.01 5481.98 M 8.49 0.06 5462.27 A 3.21 0.05 ± ± ± ± ± 5518.41 C 8.63 0.10 5469.89 M 1.53 0.04 5485.94 M 4.14 0.13 5482.48 W 8.51 0.02 5463.29 A 3.31 0.06 ± ± ± ± ± 5518.71 M 8.64 0.12 5470.38 W 1.50 0.01 5486.35 W 4.22 0.01 5482.50 C 8.47 0.15 5463.33 W 3.14 0.02 ± ± ± ± ± 5519.71 M 8.74 0.12 5471.38 W 1.51 0.01 5486.89 M 4.13 0.13 5482.59 A 8.29 0.08 5464.28 A 3.27 0.04 ± ± ± ± ± 5520.72 M 8.79 0.12 5471.49 A 1.53 0.02 5487.92 M 4.27 0.14 5482.97 M 8.46 0.06 5465.27 A 3.16 0.04 ± ± ± ± ± 5521.17 W 8.56 0.03 5472.27 W 1.50 0.01 5488.51 W 4.15 0.01 5483.50 W 8.36 0.01 5465.32 W 3.15 0.01 ± ± ± ± ± 5521.73 M 8.71 0.12 5473.30 W 1.49 0.01 5488.92 M 4.13 0.13 5483.56 C 8.60 0.15 5466.25 W 3.19 0.01 ± ± ± ± ±
(cont’d) Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5523.23 W 8.28 0.03 5474.44 W 1.49 0.01 5489.48 W 4.09 0.01 5483.61 A 8.41 0.09 5467.35 W 3.21 0.01 ± ± ± ± ± 5523.29 A 8.36 0.10 5475.39 A 1.50 0.02 5490.49 A 4.18 0.05 5484.46 W 8.42 0.01 5468.27 C 3.18 0.06 ± ± ± ± ± 5524.39 W 8.53 0.03 5476.28 W 1.48 0.01 5492.46 A 4.15 0.06 5485.41 W 8.52 0.01 5468.32 A 3.28 0.02 ± ± ± ± ± 5525.28 A 8.13 0.14 5476.86 M 1.47 0.04 5493.49 A 4.11 0.09 5486.46 W 8.45 0.01 5468.34 W 3.24 0.01 ± ± ± ± ± 5525.32 W 8.17 0.02 5477.31 W 1.44 0.01 5494.55 A 3.96 0.08 5486.97 M 8.58 0.06 5469.32 W 3.30 0.01 ± ± ± ± ±
182 5525.70 M 8.16 0.11 5477.83 M 1.43 0.03 5495.53 A 4.06 0.16 5487.98 M 8.71 0.06 5470.33 W 3.32 0.01 ± ± ± ± ± 5526.29 W 8.01 0.02 5478.22 W 1.41 0.01 5496.89 M 3.94 0.13 5488.51 W 8.59 0.01 5470.71 M 3.25 0.03 ± ± ± ± ± 5526.70 M 8.16 0.11 5479.32 W 1.44 0.01 5497.48 W 4.14 0.02 5488.96 M 8.86 0.06 5471.30 A 3.35 0.03 ± ± ± ± ± 5527.24 W 7.96 0.02 5479.84 M 1.42 0.03 5497.88 M 3.81 0.12 5489.46 W 8.70 0.02 5471.41 W 3.34 0.01 ± ± ± ± ± 5527.74 M 8.02 0.11 5481.23 W 1.44 0.01 5499.86 M 3.82 0.12 5490.42 W 8.82 0.02 5472.42 W 3.36 0.01 ± ± ± ± ± 5528.25 W 7.97 0.02 5481.87 M 1.43 0.04 5500.54 A 3.98 0.04 5492.53 A 9.13 0.10 5472.77 M 3.32 0.03 ± ± ± ± ± 5528.71 M 8.08 0.11 5482.29 W 1.43 0.01 5500.88 M 3.76 0.12 5493.52 A 9.40 0.13 5473.23 W 3.36 0.01 ± ± ± ± ± 5529.34 W 8.23 0.02 5482.40 A 1.49 0.01 5501.45 A 3.95 0.05 5493.99 M 9.35 0.07 5473.66 M 3.33 0.03 ± ± ± ± ± 5529.72 M 7.93 0.11 5482.42 C 1.43 0.03 5501.48 C 3.92 0.12 5494.57 A 9.23 0.11 5474.42 W 3.30 0.01 ± ± ± ± ± 5530.71 M 8.07 0.11 5482.85 M 1.42 0.03 5501.88 M 3.94 0.13 5495.57 A 9.27 0.15 5474.70 M 3.33 0.03 ± ± ± ± ±
(cont’d) Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5531.70 M 7.69 0.10 5483.43 A 1.43 0.02 5502.50 A 4.02 0.04 5496.95 M 9.46 0.07 5475.25 A 3.34 0.02 ± ± ± ± ± 5532.24 W 7.57 0.02 5483.48 W 1.44 0.01 5502.90 M 3.78 0.12 5497.51 W 9.38 0.02 5476.20 W 3.32 0.01 ± ± ± ± ± 5532.72 M 7.62 0.10 5483.49 C 1.46 0.03 5503.50 A 3.88 0.04 5497.94 M 9.50 0.07 5477.35 W 3.31 0.01 ± ± ± ± ± 5533.77 M 7.75 0.10 5483.85 M 1.38 0.03 5503.56 W 3.90 0.01 5498.95 M 9.54 0.07 5477.64 M 3.29 0.03 ± ± ± ± ± 5534.27 W 7.40 0.02 5484.31 W 1.43 0.01 5503.90 M 4.08 0.13 5499.93 M 9.68 0.07 5478.64 M 3.29 0.03 ± ± ± ± ±
183 5535.26 W 7.57 0.02 5484.36 A 1.52 0.04 5504.36 W 3.95 0.01 5500.49 C 9.85 0.17 5479.40 W 3.31 0.01 ± ± ± ± ± 5535.33 A 7.65 0.05 5485.31 W 1.45 0.01 5504.41 A 3.89 0.03 5500.58 A 9.71 0.11 5480.63 M 3.27 0.03 ± ± ± ± ± 5536.78 M 7.56 0.10 5485.39 A 1.47 0.02 5504.91 M 3.65 0.12 5500.95 M 9.74 0.07 5481.18 W 3.33 0.01 ± ± ± ± ± 5537.34 C 7.74 0.09 5486.29 W 1.43 0.01 5505.34 W 3.85 0.01 5501.53 A 9.61 0.12 5481.62 M 3.27 0.03 ± ± ± ± ± 5537.36 W 7.62 0.02 5486.84 M 1.45 0.04 5505.50 A 3.80 0.04 5501.55 C 9.71 0.17 5482.28 A 3.33 0.03 ± ± ± ± ± 5537.76 M 7.56 0.10 5488.50 W 1.44 0.01 5505.89 M 3.67 0.12 5501.95 M 9.75 0.07 5482.32 C 3.18 0.06 ± ± ± ± ± 5538.18 W 7.44 0.02 5489.29 W 1.37 0.01 5506.46 A 3.76 0.04 5502.55 A 9.68 0.09 5482.32 W 3.31 0.01 ± ± ± ± ± 5538.27 A 7.66 0.06 5492.41 A 1.46 0.04 5506.55 W 3.77 0.01 5502.97 M 9.70 0.07 5482.62 M 3.34 0.03 ± ± ± ± ± 5538.76 M 7.61 0.10 5493.20 W 1.42 0.02 5506.88 M 3.55 0.11 5503.53 W 9.69 0.02 5483.29 W 3.28 0.01 ± ± ± ± ± 5539.28 A 7.78 0.06 5493.37 A 1.47 0.03 5507.38 W 3.81 0.01 5503.56 A 9.70 0.14 5483.30 A 3.32 0.03 ± ± ± ± ±
(cont’d) Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5539.28 C 7.62 0.08 5494.42 A 1.51 0.03 5507.49 A 3.78 0.03 5503.97 M 9.76 0.07 5483.37 C 3.34 0.07 ± ± ± ± ± 5539.76 M 7.72 0.10 5495.23 W 1.43 0.01 5507.89 M 3.72 0.12 5504.39 W 9.70 0.02 5484.27 A 3.30 0.03 ± ± ± ± ± 5540.76 M 7.76 0.10 5495.48 A 1.49 0.06 5508.30 W 3.60 0.01 5504.55 A 9.68 0.13 5484.29 W 3.29 0.01 ± ± ± ± ± 5541.22 W 7.42 0.02 5497.37 W 1.42 0.01 5508.42 A 3.78 0.04 5504.98 M 9.69 0.07 5485.30 W 3.30 0.01 ± ± ± ± ± 5541.76 M 7.72 0.10 5497.83 M 1.43 0.03 5508.52 C 3.69 0.11 5505.35 W 9.64 0.02 5486.25 A 3.31 0.04 ± ± ± ± ±
184 5542.75 M 8.01 0.11 5498.25 W 1.42 0.01 5508.89 M 3.68 0.12 5505.54 A 9.65 0.09 5486.30 W 3.20 0.01 ± ± ± ± ± 5543.75 M 8.01 0.11 5498.36 A 1.46 0.03 5509.45 W 3.75 0.01 5505.95 M 9.68 0.07 5488.27 W 3.04 0.02 ± ± ± ± ± 5544.72 M 7.98 0.11 5498.83 M 1.44 0.04 5509.50 C 3.76 0.12 5506.48 A 9.89 0.09 5488.62 M 3.21 0.03 ± ± ± ± ± 5545.31 W 8.23 0.03 5499.81 M 1.45 0.04 5509.51 A 3.74 0.04 5506.50 W 9.68 0.02 5489.39 W 3.10 0.02 ± ± ± ± ± 5545.73 M 8.19 0.11 5500.35 A 1.46 0.01 5510.49 A 3.67 0.04 5506.95 M 9.72 0.07 5490.29 W 3.13 0.02 ± ± ± ± ± 5546.73 M 8.41 0.11 5500.83 M 1.48 0.04 5510.86 M 3.79 0.12 5507.38 W 9.65 0.02 5491.27 W 3.23 0.01 ± ± ± ± ± 5549.24 A 8.23 0.23 5501.29 A 1.44 0.01 5511.48 A 3.66 0.04 5507.54 A 9.70 0.09 5492.16 W 3.16 0.02 ± ± ± ± ± 5549.60 M 8.11 0.11 5501.43 C 1.48 0.03 5511.86 M 3.71 0.12 5507.95 M 9.73 0.07 5492.24 A 3.18 0.04 ± ± ± ± ± 5550.67 M 8.37 0.11 5501.83 M 1.43 0.04 5512.45 A 3.67 0.04 5508.55 A 9.71 0.10 5493.17 W 3.18 0.01 ± ± ± ± ± 5553.26 A 8.25 0.08 5502.22 W 1.44 0.01 5512.83 M 3.40 0.11 5508.58 C 9.48 0.16 5493.27 A 3.26 0.05 ± ± ± ± ±
(cont’d) Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5554.28 A 8.39 0.06 5502.33 A 1.48 0.01 5513.44 W 3.67 0.01 5508.96 M 9.65 0.07 5493.61 M 3.20 0.03 ± ± ± ± ± 5555.71 M 8.46 0.11 5502.85 M 1.48 0.04 5513.84 M 3.62 0.11 5509.43 W 9.62 0.02 5494.21 A 3.20 0.04 ± ± ± ± ± 5556.23 A 8.34 0.08 5503.26 W 1.46 0.01 5514.41 W 3.77 0.01 5509.54 A 9.54 0.09 5495.17 W 3.14 0.01 ± ± ± ± ± 5556.72 M 8.27 0.11 5503.34 A 1.46 0.02 5514.45 A 3.69 0.04 5509.56 C 9.72 0.17 5496.17 W 3.14 0.01 ± ± ± ± ± 5557.71 M 8.33 0.11 5503.85 M 1.42 0.03 5514.83 M 3.81 0.12 5510.54 A 9.49 0.07 5496.60 M 3.08 0.03 ± ± ± ± ±
185 5559.71 M 8.52 0.12 5504.25 W 1.46 0.01 5515.37 W 3.69 0.01 5510.93 M 9.57 0.07 5497.27 W 3.11 0.01 ± ± ± ± ± 5568.24 A 8.42 0.07 5504.27 A 1.44 0.02 5515.42 C 3.69 0.11 5511.53 A 9.43 0.10 5497.60 M 3.10 0.03 ± ± ± ± ± 5568.24 C 8.61 0.10 5504.85 M 1.41 0.03 5515.47 A 3.63 0.04 5511.92 M 9.36 0.07 5498.23 W 3.10 0.01 ± ± ± ± ± 5569.24 A 8.43 0.06 5505.23 W 1.43 0.01 5515.84 M 3.74 0.12 5512.89 M 9.30 0.07 5498.25 A 3.09 0.02 ± ± ± ± ± 5505.35 A 1.43 0.01 5516.33 W 3.55 0.01 5513.47 W 9.29 0.02 5498.61 M 3.08 0.03 ± ± ± ± 5505.83 M 1.40 0.03 5516.44 A 3.66 0.04 5513.90 M 9.23 0.07 5499.25 A 3.09 0.03 ± ± ± ± 5506.30 W 1.42 0.01 5516.49 C 3.46 0.11 5514.42 W 9.28 0.02 5499.60 M 3.03 0.03 ± ± ± ± 5506.32 A 1.44 0.02 5516.84 M 3.67 0.12 5514.47 A 9.43 0.08 5500.26 A 3.07 0.02 ± ± ± ± 5506.83 M 1.43 0.04 5517.46 A 3.68 0.04 5514.89 M 9.21 0.07 5500.29 C 3.05 0.06 ± ± ± ± 5507.33 W 1.41 0.01 5517.49 C 3.72 0.11 5515.35 W 9.23 0.02 5500.61 M 3.01 0.03 ± ± ± ± 5507.35 A 1.39 0.01 5517.84 M 3.60 0.11 5515.55 A 9.38 0.09 5501.30 C 2.99 0.06 ± ± ± ±
(cont’d) Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5507.83 M 1.37 0.03 5518.45 A 3.57 0.06 5515.57 C 9.16 0.16 5501.60 M 2.94 0.03 ± ± ± ± 5508.29 W 1.42 0.01 5518.50 C 3.52 0.11 5515.90 M 9.21 0.07 5502.19 W 2.97 0.01 ± ± ± ± 5508.34 A 1.42 0.01 5519.47 A 3.43 0.12 5516.34 W 9.31 0.02 5502.25 A 2.99 0.02 ± ± ± ± 5508.39 C 1.38 0.02 5519.53 W 3.68 0.02 5516.51 A 9.23 0.10 5502.62 M 2.95 0.03 ± ± ± ± 5508.82 M 1.42 0.03 5519.85 M 3.74 0.12 5516.53 C 9.39 0.16 5503.22 W 2.92 0.01 ± ± ± ±
186 5509.36 A 1.41 0.02 5520.45 A 3.53 0.07 5516.91 M 9.19 0.07 5503.27 A 2.96 0.04 ± ± ± ± 5509.36 C 1.38 0.02 5525.53 W 3.50 0.02 5517.54 C 9.01 0.16 5503.61 M 2.88 0.03 ± ± ± ± 5509.38 W 1.39 0.01 5525.90 M 3.66 0.12 5517.54 A 9.17 0.08 5504.25 W 2.87 0.01 ± ± ± ± 5510.81 M 1.39 0.03 5526.35 W 3.28 0.01 5517.91 M 9.12 0.07 5504.31 A 2.85 0.03 ± ± ± ± 5511.21 W 1.36 0.01 5526.52 C 3.24 0.10 5518.54 A 9.08 0.10 5504.62 M 2.87 0.03 ± ± ± ± 5511.32 A 1.38 0.02 5526.83 M 3.57 0.11 5518.56 C 9.42 0.16 5505.28 A 2.83 0.02 ± ± ± ± 5511.80 M 1.36 0.03 5527.45 W 3.46 0.01 5518.90 M 9.20 0.07 5505.31 W 2.84 0.01 ± ± ± ± 5512.33 A 1.38 0.02 5527.85 M 3.59 0.11 5519.48 W 9.28 0.02 5505.61 M 2.83 0.03 ± ± ± ± 5512.75 M 1.31 0.03 5528.26 W 3.33 0.01 5519.91 M 9.21 0.07 5506.23 A 2.86 0.02 ± ± ± ± 5513.40 W 1.36 0.01 5528.84 M 3.43 0.11 5520.50 A 9.10 0.11 5506.25 W 2.83 0.01 ± ± ± ±
(cont’d) Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5513.78 M 1.42 0.03 5529.84 M 3.48 0.11 5521.49 W 9.32 0.02 5506.61 M 2.83 0.03 ± ± ± ± 5514.30 A 1.35 0.02 5530.31 W 3.38 0.01 5523.90 M 9.43 0.07 5507.24 A 2.85 0.02 ± ± ± ± 5514.41 W 1.36 0.01 5530.84 M 3.60 0.11 5524.40 W 9.52 0.03 5507.29 W 2.84 0.01 ± ± ± ± 5514.78 M 1.40 0.03 5531.29 W 3.29 0.01 5525.44 W 9.34 0.02 5507.62 M 2.76 0.03 ± ± ± ± 5515.33 A 1.30 0.02 5531.84 M 3.30 0.10 5526.34 W 9.36 0.02 5508.16 W 2.81 0.01 ± ± ± ±
187 5515.33 C 1.33 0.02 5532.32 W 3.35 0.01 5526.59 C 9.63 0.17 5508.25 A 2.79 0.03 ± ± ± ± 5515.35 W 1.25 0.01 5532.83 M 3.20 0.10 5526.89 M 9.57 0.07 5508.33 C 2.80 0.06 ± ± ± ± 5515.79 M 1.36 0.03 5533.85 M 3.19 0.10 5527.41 W 9.47 0.02 5508.61 M 2.76 0.03 ± ± ± ± 5516.34 A 1.34 0.02 5534.28 W 3.31 0.01 5527.91 M 9.63 0.07 5509.22 A 2.85 0.02 ± ± ± ± 5517.34 A 1.35 0.05 5535.28 W 3.26 0.01 5528.91 M 9.73 0.07 5509.23 C 2.88 0.06 ± ± ± ± 5520.34 A 1.28 0.03 5535.41 A 3.21 0.04 5529.91 M 9.78 0.07 5509.27 W 2.83 0.01 ± ± ± ± 5523.23 W 1.26 0.01 5535.82 M 3.11 0.10 5530.32 W 9.98 0.02 5509.62 M 2.79 0.03 ± ± ± ± 5524.39 W 1.30 0.01 5536.84 M 3.11 0.10 5530.90 M 9.90 0.07 5510.30 A 2.86 0.04 ± ± ± ± 5525.34 W 1.27 0.01 5537.44 W 3.19 0.01 5531.29 W 9.85 0.02 5510.62 M 2.84 0.03 ± ± ± ± 5526.29 W 1.26 0.01 5537.83 M 3.06 0.10 5531.90 M 10.04 0.07 5511.16 W 2.81 0.01 ± ± ± ±
(cont’d) Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5526.79 M 1.25 0.03 5538.41 A 3.16 0.03 5532.44 W 9.98 0.02 5511.18 A 2.84 0.03 ± ± ± ± 5527.26 W 1.26 0.01 5538.83 M 3.07 0.10 5532.89 M 10.02 0.07 5511.62 M 2.82 0.03 ± ± ± ± 5527.78 M 1.34 0.03 5539.36 C 3.17 0.10 5533.91 M 10.07 0.07 5512.26 A 2.85 0.03 ± ± ± ± 5528.25 W 1.27 0.01 5539.39 A 3.12 0.03 5534.32 W 10.13 0.02 5512.62 M 2.88 0.03 ± ± ± ± 5528.79 M 1.40 0.03 5539.83 M 3.03 0.10 5534.50 A 10.17 0.09 5513.22 A 2.94 0.05 ± ± ± ±
188 5529.77 M 1.30 0.03 5540.84 M 3.05 0.10 5535.52 A 10.15 0.07 5513.61 M 2.96 0.03 ± ± ± ± 5530.78 M 1.25 0.03 5541.26 W 3.13 0.01 5535.88 M 10.11 0.07 5514.21 A 2.93 0.03 ± ± ± ± 5531.22 W 1.25 0.01 5541.83 M 3.05 0.10 5536.91 M 10.30 0.08 5514.61 M 2.94 0.03 ± ± ± ± 5531.77 M 1.27 0.03 5542.86 M 3.18 0.10 5537.46 W 10.23 0.02 5515.19 C 2.97 0.06 ± ± ± ± 5532.36 W 1.25 0.01 5543.38 A 3.10 0.04 5537.90 M 10.25 0.08 5515.21 A 3.01 0.03 ± ± ± ± 5532.77 M 1.24 0.03 5543.82 M 3.10 0.10 5538.47 C 10.26 0.18 5515.62 M 2.95 0.03 ± ± ± ± 5533.73 M 1.30 0.03 5544.80 M 3.11 0.10 5538.50 A 10.29 0.10 5516.19 A 3.04 0.03 ± ± ± ± 5534.25 A 1.27 0.02 5545.34 W 3.20 0.01 5538.89 M 10.23 0.08 5516.26 C 3.01 0.06 ± ± ± ± 5534.27 W 1.26 0.01 5545.82 M 3.04 0.10 5539.46 C 10.53 0.18 5516.61 M 2.97 0.03 ± ± ± ± 5535.30 W 1.27 0.01 5546.84 M 3.30 0.10 5539.51 A 10.23 0.10 5517.26 A 2.93 0.04 ± ± ± ±
(cont’d) Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5535.75 M 1.32 0.03 5549.40 A 3.04 0.09 5539.89 M 10.23 0.08 5517.30 C 3.10 0.06 ± ± ± ± 5536.73 M 1.28 0.03 5549.79 M 3.15 0.10 5540.90 M 10.27 0.08 5517.61 M 2.93 0.01 ± ± ± ± 5537.37 W 1.27 0.01 5553.41 A 3.24 0.06 5541.89 M 10.31 0.08 5518.20 W 2.95 0.04 ± ± ± ± 5537.67 M 1.26 0.03 5554.38 A 3.15 0.04 5542.89 M 10.41 0.08 5518.21 A 3.12 0.06 ± ± ± ± 5538.18 W 1.32 0.01 5555.30 A 3.11 0.07 5543.44 A 10.53 0.10 5518.31 C 2.97 0.03 ± ± ± ±
189 5538.29 A 1.26 0.01 5557.79 M 3.40 0.11 5543.89 M 10.44 0.08 5519.27 A 2.94 0.10 ± ± ± ± 5538.32 C 1.32 0.02 5559.78 M 3.35 0.11 5544.86 M 10.42 0.08 5519.61 M 3.02 0.03 ± ± ± ± 5538.71 M 1.28 0.03 5566.37 A 3.70 0.03 5545.32 W 10.66 0.02 5520.19 W 3.00 0.01 ± ± ± ± 5539.30 A 1.28 0.01 5567.31 A 3.76 0.03 5545.89 M 10.57 0.08 5520.24 A 2.98 0.03 ± ± ± ± 5539.31 C 1.30 0.02 5568.35 A 3.75 0.03 5546.94 M 10.58 0.08 5521.15 W 3.01 0.01 ± ± ± ± 5539.67 M 1.26 0.03 5569.32 C 3.73 0.11 5549.49 A 10.83 0.14 5523.16 W 3.01 0.01 ± ± ± ± 5540.72 M 1.31 0.03 5549.86M10.73 0.08 5523.20 A 2.97 0.05 ± ± ± 5541.25 W 1.29 0.01 5550.47A10.84 0.12 5525.20 A 2.98 0.03 ± ± ± 5541.71 M 1.35 0.03 5550.85M10.89 0.08 5525.24 W 2.97 0.01 ± ± ± 5542.70 M 1.34 0.03 5553.44A10.70 0.16 5525.24 C 2.94 0.06 ± ± ±
(cont’d) Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5543.70 M 1.40 0.03 5554.43A11.04 0.11 5525.60 M 2.99 0.03 ± ± ± 5546.68 M 1.43 0.03 5555.32A11.62 0.29 5526.24 W 2.99 0.01 ± ± ± 5549.67 M 1.47 0.04 5555.83M11.13 0.08 5526.60 M 2.96 0.03 ± ± ± 5550.67 M 1.43 0.04 5556.85M10.94 0.08 5527.22 W 2.94 0.01 ± ± ± 5553.31 A 1.35 0.03 5559.86M10.89 0.08 5527.63 M 2.98 0.03 ± ± ±
190 5554.32 A 1.39 0.03 5566.42A11.04 0.09 5528.24 W 2.93 0.01 ± ± ± 5555.26 A 1.39 0.03 5567.46A11.10 0.08 5528.61 M 2.94 0.03 ± ± ± 5556.67 M 1.39 0.03 5569.39A10.91 0.10 5529.61 M 2.95 0.03 ± ± ± 5559.67 M 1.39 0.03 5530.61 M 2.94 0.03 ± ± 5568.21 C 1.41 0.02 5531.18 W 2.89 0.01 ± ± 5568.27 A 1.39 0.01 5531.60 M 2.95 0.03 ± ± 5532.20 W 2.89 0.01 ± 5532.61 M 2.95 0.03 ± 5533.61 M 2.93 0.03 ± 5534.13 A 2.86 0.06 ±
(cont’d) Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5534.20 W 2.89 0.01 ± 5535.17 W 2.88 0.01 ± 5536.61 M 2.90 0.03 ± 5538.16 W 2.85 0.01 ± 5538.17 C 2.96 0.06 ±
191 5538.20 A 2.82 0.02 ± 5538.61 M 2.87 0.03 ± 5539.20 A 2.85 0.02 ± 5539.22 C 2.87 0.06 ± 5539.61 M 2.94 0.03 ± 5540.60 M 2.90 0.03 ± 5541.21 W 2.79 0.01 ± 5541.57 M 2.92 0.03 ± 5542.59 M 2.83 0.03 ± 5543.60 M 2.92 0.03 ±
(cont’d) Table A.1—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099 a b a b a b a b a b HJD Fcon HJD Fcon HJD Fcon HJD Fcon HJD Fcon
5544.61 M 2.86 0.03 ± 5548.61 M 2.86 0.03 ± 5549.19 A 2.84 0.04 ± 192 5551.18 A 2.93 0.09 ± 5553.20 A 2.90 0.05 ± 5555.20 A 2.91 0.04 ± 5555.60 M 3.17 0.03 ± 5556.19 A 2.96 0.14 ± 5557.60 M 2.99 0.03 ±
Note. — Dates are given as Heliocentric Julian Date ( 2450000). Continuum fluxes are in units of 10−15 ergs s−1 cm−2 − A˚−1. Observatory Codes: C=CRAO Spectroscopy, A=CRAO Photometry, W= WISE, M=MDM Mrk335 Mrk1501 3C120 Mrk6 PG2130+099
HJD FHβ HJD FHβ HJD FHβ HJD FHβ HJD FHβ
5440.9 M 5.32 0.08 5440.9 M 2.45 0.06 5441.9 M 3.36 0.07 5442.0 M 6.21 0.09 5441.7 M 4.38 0.05 ± ± ± ± ± 5441.8 M 5.49 0.08 5441.8 M 2.54 0.07 5443.9 M 3.50 0.07 5442.9 M 6.12 0.09 5442.7 M 4.33 0.05 ± ± ± ± ± 5442.8 M 5.29 0.08 5442.8 M 2.53 0.07 5446.9 M 3.43 0.07 5443.5 C 6.38 0.22 5443.4 C 4.37 0.07 ± ± ± ± ± 5444.8 M 5.28 0.08 5443.5 C 2.40 0.07 5447.9 M 3.37 0.07 5444.9 M 6.14 0.09 5443.7 M 4.30 0.05 ± ± ± ± ± 5445.8 M 5.21 0.08 5444.4 C 2.45 0.08 5452.9 M 3.30 0.07 5450.5 C 6.21 0.22 5444.3 C 4.29 0.07 ± ± ± ± ± 5447.9 M 5.26 0.08 5444.8 M 2.51 0.06 5454.9 M 3.30 0.07 5450.9 M 6.01 0.09 5444.7 M 4.39 0.05 ± ± ± ± ± 5449.9 M 5.56 0.08 5446.9 M 2.50 0.06 5455.8 M 3.37 0.07 5451.4 C 6.11 0.21 5445.7 M 4.39 0.05 ± ± ± ± ±
193 5451.7 M 5.47 0.08 5449.8 M 2.58 0.07 5456.5 C 3.31 0.14 5453.9 M 6.04 0.09 5446.7 M 4.30 0.05 ± ± ± ± ± 5453.8 M 5.45 0.08 5450.8 M 2.52 0.06 5457.4 C 3.39 0.15 5454.9 M 5.82 0.09 5447.6 M 4.34 0.05 ± ± ± ± ± 5454.7 M 5.44 0.08 5451.4 C 2.45 0.08 5457.9 M 3.41 0.07 5455.9 M 5.93 0.09 5449.7 M 4.36 0.05 ± ± ± ± ± 5456.8 M 5.36 0.08 5451.8 M 2.57 0.07 5458.5 C 3.59 0.15 5456.4 C 6.02 0.21 5450.7 M 4.31 0.05 ± ± ± ± ± 5457.7 M 5.50 0.08 5452.4 C 2.52 0.08 5458.9 M 3.38 0.07 5456.9 M 5.99 0.09 5451.3 C 4.31 0.07 ± ± ± ± ± 5458.8 M 5.32 0.08 5453.8 M 2.52 0.06 5459.5 C 3.44 0.15 5457.4 C 6.06 0.21 5451.6 M 4.31 0.05 ± ± ± ± ± 5466.8 M 5.12 0.08 5454.3 C 2.41 0.07 5466.8 M 3.60 0.08 5457.9 M 5.88 0.09 5452.3 C 4.39 0.07 ± ± ± ± ± 5467.8 M 5.20 0.08 5454.8 M 2.44 0.06 5470.9 M 3.55 0.08 5458.5 C 6.02 0.21 5453.6 M 4.40 0.05 ± ± ± ± ± 5468.8 M 4.97 0.07 5455.8 M 2.45 0.06 5471.9 M 3.49 0.07 5459.4 C 6.08 0.21 5454.3 C 4.35 0.07 ± ± ± ± ±
(cont’d) Table A.2. Hβ Fluxes Table A.2—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099
HJD FHβ HJD FHβ HJD FHβ HJD FHβ HJD FHβ
5469.8 M 5.06 0.07 5456.3 C 2.44 0.08 5472.9 M 3.59 0.08 5463.9 M 6.04 0.09 5454.6 M 4.37 0.05 ± ± ± ± ± 5470.9 M 5.08 0.08 5457.3 C 2.46 0.08 5476.9 M 3.50 0.07 5466.9 M 6.08 0.09 5456.2 C 4.21 0.06 ± ± ± ± ± 5472.9 M 5.04 0.07 5457.8 M 2.37 0.06 5477.9 M 3.41 0.07 5467.9 M 5.94 0.09 5456.6 M 4.34 0.05 ± ± ± ± ± 5473.8 M 4.97 0.07 5458.4 C 2.44 0.08 5479.8 M 3.54 0.07 5468.4 C 6.17 0.22 5457.2 C 4.27 0.07 ± ± ± ± ± 5476.7 M 5.21 0.08 5458.8 M 2.41 0.06 5480.8 M 3.46 0.07 5468.9 M 5.94 0.09 5457.6 M 4.23 0.05 ± ± ± ± ± 5477.7 M 5.08 0.08 5468.3 C 2.30 0.07 5481.9 M 3.78 0.08 5471.9 M 6.08 0.09 5468.2 C 4.31 0.07 ± ± ± ± ±
194 5478.9 M 5.03 0.07 5468.8 M 2.49 0.06 5482.5 C 3.58 0.15 5476.9 M 6.28 0.09 5470.7 M 4.33 0.05 ± ± ± ± ± 5479.7 M 5.06 0.07 5469.8 M 2.44 0.06 5482.9 M 3.56 0.08 5477.9 M 6.42 0.09 5472.7 M 4.27 0.05 ± ± ± ± ± 5480.8 M 5.06 0.08 5476.8 M 2.48 0.06 5483.5 C 3.58 0.15 5478.9 M 6.27 0.09 5473.6 M 4.25 0.05 ± ± ± ± ± 5481.8 M 5.07 0.08 5477.8 M 2.29 0.06 5483.8 M 3.54 0.07 5479.9 M 6.35 0.09 5474.7 M 4.32 0.05 ± ± ± ± ± 5482.7 M 5.23 0.08 5479.8 M 2.23 0.06 5485.9 M 3.68 0.08 5480.9 M 6.27 0.09 5477.6 M 4.27 0.05 ± ± ± ± ± 5483.7 M 5.09 0.08 5481.8 M 2.31 0.06 5486.8 M 3.65 0.08 5481.9 M 6.16 0.09 5478.6 M 4.30 0.05 ± ± ± ± ± 5486.7 M 5.26 0.08 5482.4 C 2.24 0.07 5487.9 M 3.71 0.08 5482.5 C 6.24 0.22 5480.6 M 4.28 0.05 ± ± ± ± ± 5488.7 M 5.24 0.08 5482.8 M 2.18 0.06 5488.9 M 3.70 0.08 5482.9 M 6.31 0.09 5481.6 M 4.29 0.05 ± ± ± ± ± 5497.7 M 5.27 0.08 5483.4 C 2.09 0.06 5496.8 M 3.83 0.08 5483.5 C 6.46 0.23 5482.3 C 4.36 0.07 ± ± ± ± ± 5498.7 M 5.30 0.08 5483.8 M 2.21 0.06 5497.8 M 3.73 0.08 5486.9 M 6.51 0.10 5482.6 M 4.32 0.05 ± ± ± ± ±
(cont’d) Table A.2—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099
HJD FHβ HJD FHβ HJD FHβ HJD FHβ HJD FHβ
5499.7 M 5.32 0.08 5486.8 M 2.26 0.06 5499.8 M 3.81 0.08 5487.9 M 6.39 0.09 5483.3 C 4.28 0.07 ± ± ± ± ± 5500.7 M 5.38 0.08 5497.7 M 2.09 0.05 5500.8 M 3.71 0.08 5488.9 M 6.72 0.10 5488.6 M 4.22 0.05 ± ± ± ± ± 5501.7 M 5.25 0.08 5498.8 M 2.18 0.06 5501.4 C 3.85 0.17 5493.9 M 6.55 0.10 5493.6 M 4.33 0.05 ± ± ± ± ± 5502.7 M 5.25 0.08 5499.7 M 2.05 0.05 5501.8 M 3.84 0.08 5496.9 M 6.65 0.10 5496.6 M 4.33 0.05 ± ± ± ± ± 5503.7 M 5.32 0.08 5500.8 M 2.21 0.06 5502.9 M 3.79 0.08 5497.9 M 6.95 0.10 5497.5 M 4.23 0.05 ± ± ± ± ± 5504.7 M 5.30 0.08 5501.4 C 2.15 0.07 5503.9 M 3.93 0.08 5498.9 M 6.95 0.10 5498.6 M 4.31 0.05 ± ± ± ± ±
195 5505.7 M 5.35 0.08 5501.7 M 2.06 0.05 5504.9 M 3.76 0.08 5499.9 M 6.89 0.10 5499.5 M 4.34 0.05 ± ± ± ± ± 5506.7 M 5.40 0.08 5502.8 M 2.06 0.05 5505.8 M 3.70 0.08 5500.4 C 6.69 0.24 5500.2 C 4.20 0.06 ± ± ± ± ± 5507.7 M 5.38 0.08 5503.8 M 2.04 0.05 5506.8 M 3.79 0.08 5500.9 M 7.13 0.11 5500.6 M 4.21 0.05 ± ± ± ± ± 5508.7 M 5.39 0.08 5504.8 M 2.07 0.05 5507.8 M 3.89 0.08 5501.5 C 7.13 0.25 5501.2 C 4.12 0.06 ± ± ± ± ± 5509.4 C 5.66 0.09 5505.8 M 2.10 0.05 5508.5 C 4.11 0.18 5501.9 M 7.02 0.10 5501.6 M 4.27 0.05 ± ± ± ± ± 5510.7 M 5.66 0.08 5506.7 M 2.09 0.05 5508.8 M 3.96 0.08 5502.9 M 7.18 0.11 5502.6 M 4.32 0.05 ± ± ± ± ± 5513.7 M 5.48 0.08 5507.8 M 2.06 0.05 5509.5 C 4.07 0.18 5503.9 M 7.08 0.10 5503.6 M 4.29 0.05 ± ± ± ± ± 5514.7 M 5.67 0.08 5508.3 C 2.12 0.07 5510.8 M 4.13 0.09 5504.9 M 7.22 0.11 5504.6 M 4.27 0.05 ± ± ± ± ± 5515.3 C 5.84 0.09 5508.7 M 2.05 0.05 5511.8 M 4.01 0.09 5505.9 M 7.12 0.11 5505.6 M 4.29 0.05 ± ± ± ± ± 5515.7 M 5.77 0.09 5509.3 C 2.11 0.06 5512.8 M 3.88 0.08 5506.9 M 7.36 0.11 5506.6 M 4.24 0.05 ± ± ± ± ±
(cont’d) Table A.2—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099
HJD FHβ HJD FHβ HJD FHβ HJD FHβ HJD FHβ
5516.7 M 5.90 0.09 5510.7 M 2.01 0.05 5513.8 M 4.02 0.09 5507.9 M 7.33 0.11 5507.6 M 4.20 0.05 ± ± ± ± ± 5517.3 C 5.74 0.09 5511.7 M 2.01 0.05 5514.8 M 4.10 0.09 5508.5 C 7.50 0.26 5508.3 C 4.15 0.06 ± ± ± ± ± 5517.7 M 5.93 0.09 5512.7 M 2.07 0.05 5515.4 C 3.98 0.17 5508.9 M 7.23 0.11 5508.6 M 4.25 0.05 ± ± ± ± ± 5518.4 C 5.66 0.09 5513.7 M 2.09 0.05 5515.8 M 4.10 0.09 5509.5 C 7.39 0.26 5509.2 C 4.18 0.06 ± ± ± ± ± 5518.7 M 5.73 0.09 5514.7 M 2.20 0.06 5516.4 C 3.82 0.16 5510.9 M 7.34 0.11 5509.6 M 4.25 0.05 ± ± ± ± ± 5519.7 M 6.09 0.09 5515.3 C 2.15 0.07 5516.8 M 4.03 0.09 5511.9 M 7.29 0.11 5510.6 M 4.17 0.05 ± ± ± ± ±
196 5520.7 M 6.03 0.09 5515.7 M 2.11 0.05 5517.4 C 4.06 0.18 5512.8 M 7.16 0.11 5511.6 M 4.17 0.05 ± ± ± ± ± 5521.7 M 6.24 0.09 5526.7 M 1.87 0.05 5517.8 M 3.99 0.08 5513.9 M 7.38 0.11 5512.6 M 4.08 0.05 ± ± ± ± ± 5525.7 M 6.30 0.09 5527.7 M 1.93 0.05 5518.5 C 3.82 0.16 5514.8 M 7.42 0.11 5513.6 M 3.94 0.05 ± ± ± ± ± 5526.7 M 6.51 0.10 5528.7 M 1.96 0.05 5519.8 M 4.16 0.09 5515.5 C 7.34 0.26 5514.6 M 4.01 0.05 ± ± ± ± ± 5527.7 M 6.55 0.10 5529.7 M 1.91 0.05 5525.9 M 4.11 0.09 5515.9 M 7.28 0.11 5515.1 C 4.05 0.06 ± ± ± ± ± 5528.7 M 6.52 0.10 5530.7 M 1.94 0.05 5526.5 C 3.81 0.16 5516.5 C 7.48 0.26 5515.6 M 4.07 0.05 ± ± ± ± ± 5529.7 M 6.64 0.10 5531.7 M 1.94 0.05 5526.8 M 4.22 0.09 5516.9 M 7.42 0.11 5516.2 C 4.07 0.06 ± ± ± ± ± 5530.7 M 6.69 0.10 5532.7 M 1.90 0.05 5527.8 M 4.19 0.09 5517.5 C 7.50 0.26 5516.6 M 3.97 0.05 ± ± ± ± ± 5531.7 M 6.54 0.10 5533.6 M 2.01 0.05 5528.8 M 3.96 0.08 5517.9 M 7.35 0.11 5517.3 C 3.98 0.06 ± ± ± ± ± 5532.7 M 6.61 0.10 5535.7 M 1.83 0.05 5529.8 M 4.07 0.09 5518.5 C 7.54 0.27 5517.6 M 4.02 0.06 ± ± ± ± ±
(cont’d) Table A.2—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099
HJD FHβ HJD FHβ HJD FHβ HJD FHβ HJD FHβ
5533.7 M 6.62 0.10 5536.7 M 1.98 0.05 5530.8 M 4.11 0.09 5518.9 M 7.28 0.11 5518.3 C 3.89 0.05 ± ± ± ± ± 5536.7 M 6.38 0.10 5537.6 M 1.92 0.05 5531.8 M 3.86 0.08 5519.9 M 7.36 0.11 5519.6 M 3.92 0.05 ± ± ± ± ± 5537.3 C 6.41 0.10 5538.3 C 1.88 0.06 5532.8 M 3.77 0.08 5523.9 M 7.24 0.11 5525.2 C 4.15 0.06 ± ± ± ± ± 5537.7 M 6.37 0.10 5538.6 M 1.89 0.05 5533.8 M 3.81 0.08 5526.5 C 7.48 0.26 5525.5 M 3.96 0.05 ± ± ± ± ± 5538.7 M 6.20 0.09 5539.3 C 1.97 0.06 5535.8 M 3.82 0.08 5526.8 M 7.42 0.11 5526.6 M 3.94 0.05 ± ± ± ± ± 5539.2 C 6.56 0.10 5539.6 M 1.86 0.05 5536.8 M 3.73 0.08 5527.9 M 7.30 0.11 5527.6 M 4.05 0.05 ± ± ± ± ±
197 5539.7 M 6.38 0.10 5540.6 M 1.93 0.05 5537.8 M 3.79 0.08 5528.9 M 7.03 0.10 5528.6 M 3.98 0.05 ± ± ± ± ± 5540.7 M 6.28 0.09 5541.6 M 1.98 0.05 5538.8 M 3.80 0.08 5529.9 M 7.30 0.11 5529.6 M 4.13 0.05 ± ± ± ± ± 5541.7 M 6.19 0.09 5542.6 M 1.87 0.05 5539.3 C 4.07 0.18 5530.9 M 7.17 0.11 5530.6 M 4.05 0.05 ± ± ± ± ± 5542.7 M 6.37 0.10 5543.6 M 1.79 0.04 5539.8 M 3.81 0.08 5531.9 M 7.35 0.11 5531.6 M 4.07 0.05 ± ± ± ± ± 5543.7 M 6.13 0.09 5546.6 M 1.95 0.05 5540.8 M 3.78 0.08 5532.8 M 7.43 0.11 5532.6 M 4.03 0.05 ± ± ± ± ± 5544.7 M 6.26 0.09 5549.7 M 1.95 0.05 5541.8 M 3.76 0.08 5533.9 M 7.24 0.11 5533.6 M 4.06 0.05 ± ± ± ± ± 5545.7 M 6.17 0.09 5550.7 M 1.89 0.05 5542.8 M 3.75 0.08 5535.8 M 7.46 0.11 5536.6 M 4.01 0.05 ± ± ± ± ± 5546.7 M 6.29 0.09 5556.6 M 1.87 0.05 5543.8 M 3.74 0.08 5536.9 M 7.53 0.11 5538.1 C 3.92 0.06 ± ± ± ± ± 5549.6 M 6.20 0.09 5559.6 M 1.87 0.05 5544.8 M 3.69 0.08 5537.9 M 7.82 0.12 5538.6 M 3.86 0.05 ± ± ± ± ± 5550.6 M 6.25 0.09 5568.2 C 1.98 0.06 5545.8 M 3.69 0.08 5538.4 C 7.48 0.26 5539.2 C 4.02 0.06 ± ± ± ± ±
(cont’d) Table A.2—Continued
Mrk335 Mrk1501 3C120 Mrk6 PG2130+099
HJD FHβ HJD FHβ HJD FHβ HJD FHβ HJD FHβ
5555.7 M 6.30 0.09 5546.8M3.89 0.08 5538.8 M 8.05 0.12 5539.6 M 3.90 0.05 ± ± ± ± 5556.7 M 6.43 0.10 5549.7M3.69 0.08 5539.4 C 7.45 0.26 5540.6 M 3.97 0.05 ± ± ± ± 5557.7 M 6.51 0.10 5557.7M3.69 0.08 5539.8 M 7.98 0.12 5541.5 M 3.87 0.05 ± ± ± ± 5559.7 M 6.49 0.10 5559.7M3.54 0.07 5540.9 M 7.88 0.12 5542.5 M 4.00 0.05 ± ± ± ± 5568.2 C 6.57 0.10 5569.3C3.39 0.15 5541.8 M 7.85 0.12 5543.5 M 3.84 0.05 ± ± ± ±
198 5542.8 M 8.04 0.12 5544.6 M 3.89 0.05 ± ± 5543.8 M 8.07 0.12 5548.6 M 3.87 0.05 ± ± 5544.8 M 8.10 0.12 5555.6 M 3.92 0.05 ± ± 5545.8 M 7.95 0.12 5557.5 M 3.89 0.05 ± ± 5546.9 M 8.11 0.12 ± 5549.8 M 8.15 0.12 ± 5550.8 M 8.11 0.12 ± 5555.8 M 7.9 0.12 ± 5556.8 M 8.0 0.12 ± 5559.8 M 8.0 0.12 ±
Note. — Dates are in Heliocentric Julian Date ( 2450000). Hβ flux is in units of 10−13 ergs s−1cm−2. Observatory − Codes: M=MDM Observatory, C=CrAO