The Pennsylvania State University
The Graduate School
College of Science
BROAD ABSORPTION LINE VARIABILITY
ON MULTI-YEAR TIMESCALES
IN A LARGE QUASAR SAMPLE
A Dissertation in
Astronomy and Astrophysics
by
Nurten Filiz Ak
c 2014 Nurten Filiz Ak
Submitted in Partial Fulfillment of the Requirements for the Degree of
Doctor of Philosophy
May 2014 The dissertation of Nurten Filiz Ak was reviewed and approved∗ by the following:
William Nielsen Brandt Distinguished Professor of Astronomy and Astrophysics Dissertation Adviser Chair of Committee
Donald Schneider Distinguished Professor of Astronomy and Astrophysics Head of the Department of Astronomy and Astrophysics
Michael Eracleous Professor of Astronomy and Astrophysics
Mercedes Richards Professor of Astronomy and Astrophysics
Benjamin Owen Professor of Physics
Steinn Sigurdsson Professor of Astronomy and Astrophysics Chair of the Graduate Program of Astronomy and Astrophysics
∗Signatures on file in the Graduate School. iii Abstract
Outflows launched near the central supermassive black holes (SMBHs) are a common and important component of active galactic nuclei (AGNs). Outflows in luminous AGNs (i.e., quasars) play a key role in mass accretion onto SMBH as well as in the feedback into host galaxies. The most prominent signature of such outflows appears as broad absorption lines (BALs) that are blueshifted from the emission line with a few thousands kms−1 velocities. In this dissertation, I place further constrains upon the size scale, internal structure, dynamics, and evolution of the outflows investigating profiles, properties, and variation characteristics of BAL troughs. I present observational results on BAL troughs in a large quasar sample utilizing spectroscopic observations from the Sloan Digital Sky Survey spanning on multi-year timescales. The results presented here, for the first time, provide a large and well-defined variability data base capable of discriminating between time-dependent hydrodynamic wind calculations in a statistically powerful manner. In a study of 582 quasars, I present 21 examples of BAL trough disappearance. Ap- proximately 3.3% of BAL quasars show disappearing C iv trough on rest-frame timescales of 1.1–3.9 yr. BAL disappearance appears to occur mainly for shallow and weak or moderate- strength absorption troughs but not the strongest ones. When one BAL trough in a quasar spectrum disappears, the other present troughs usually weaken. Possible causes of such coordinated variations could be disk-wind rotation or variations of shielding gas that lead to variations of ionizing-continuum radiation. I present a detailed study on the variability of 428 C iv and 235 Si iv BAL troughs using a systematically observed sample of 291 BAL quasars. BAL variation distributions indicate that BAL disappearance is an extreme type of general BAL variability, rather than a qualitatively distinct phenomenon. The high observed frequency of BAL variability on multi-year timescales is generally supportive of models where most BAL absorption arises at radii of 10–1000 light days. Average lifetime for a BAL trough along our line-of-sight is a few thousand years which is long compared to the orbital time of the accretion disk at the wind-launching radius. We have examined if BAL variations on several timescales depend upon quasar properties, including quasar luminosity, Eddington luminosity ratio, black hole mass, redshift, and radio loudness. Within the ranges of these properties spanned by our sample, we do not find any strong dependences. The coordinated trough variability of BAL quasars with multiple troughs suggests that changes in ”shielding gas” may play a significant role in driving general BAL variability. I present a study investigating the dependence of C iv BAL properties and variation characteristics on accompanying Si iv and Al iii absorption. Results of this study show that C iv BAL trough shapes, depths, velocity widths and strengths show a strong dependence on the presence of Si iv and Al iii BAL troughs at corresponding velocities. Similarly, the variation characteristics and depth variation profiles of C iv BAL troughs also show a iv strong connection to BAL troughs in these transitions. Using these ions as a basic tracer of ionization level of the absorbing gas, systematic measurements of variability and profiles for a large sample of C iv, Si iv, and Al iii BAL troughs present observational evidences of the relation between ionization level, column density and kinematics of outflows. Utilizing observational investigations on a large BAL quasar sample, we show that ionization level, column density and kinematics of outflows show correlated object-to-object differences. We present a detailed comparison between the observational results of this study and the well studied disk-wind model of quasar outflows, which suggests that the wind is launched from the accretion disk at 1016–1017 cm and radiatively driven by ∼ UV line pressure. Results of this study show that lines-of-sight with different viewing inclinations successfully explain the characteristics and the differences between those three C iv trough groups with a good agreement to our observational findings. v Table of Contents
List of Tables ...... viii
List of Figures ...... ix
Acknowledgments ...... xii
Chapter 1. Introduction ...... 1 1.1 Active Galactic Nuclei ...... 1 1.2 BroadAbsorptionLineQuasars ...... 1 1.3 A Large-Scale Survey of Quasar BAL Variability with SDSS BOSS.. 3 1.4 OutlineoftheThesis...... 3
Chapter 2. Broad Absorption Line Disappearance on Multi-Year Timescales in a Large Quasar Sample ...... 6 2.1 Introduction...... 6 2.2 Observations and Sample Selection ...... 8 2.2.1 Observations ...... 8 2.2.2 SampleSelection ...... 9 2.3 Identification of Disappearing BALs ...... 11 2.3.1 Continuum Fit and Normalization ...... 11 2.3.2 Measurements of BAL Properties ...... 13 2.3.3 Selection of Disappearing BALs ...... 15 2.3.4 Notes on Specific Objects ...... 22 2.4 Statistical Properties of Disappearing BALs ...... 26 2.4.1 How Common is BAL Disappearance? ...... 26 2.4.2 Luminosities, Black-Hole Masses, Reddening, and Radio Prop- erties of Quasars with Disappearing BAL Troughs ...... 27 2.4.3 EWs, Depths, Velocities, and Widths of Disappearing BAL Troughs 30 2.4.4 Connections Between Disappearing and Non-Disappearing Troughs 32 2.5 SummaryandFutureWork ...... 36
Chapter 3. Broad Absorption Line Variation on Multi-Year Timescales in a Large Quasar Sample ...... 43 3.1 Introduction...... 43 3.2 Observations and Sample Selection ...... 45 3.2.1 Observations ...... 45 3.2.2 SampleSelection ...... 46 vi
3.3 DataPreparationandAnalysis ...... 47 3.3.1 Basic Spectral Preparation ...... 47 3.3.2 Identification and Measurements of BAL Troughs ...... 48 3.3.3 Identification and Measurements of Variable BAL Troughs... 54 3.4 ResultsonBALVariability ...... 55 3.4.1 Fraction of BAL Troughs and BAL Quasars Showing Variability 56 3.4.2 Velocity Widths of Variable Regions of BALs ...... 59 3.4.3 EW Variations as a Function of Timescale ...... 61 3.4.4 Distribution of EW Variations ...... 66 3.4.5 EW Variations as a Function of BAL-Profile Properties . . . . 70 3.4.6 Comparison of C iv vs. Si iv EW Variations ...... 75 3.4.7 Coordination of EW Variations in BAL Quasars with Multiple Troughs ...... 79 3.4.8 EW Variations as a Function of Quasar Properties ...... 83 3.4.8.1 Luminosity ...... 83 3.4.8.2 Eddington Luminosity Ratio and SMBH Mass . . . . 85 3.4.8.3 Redshift...... 87 3.4.8.4 RadioLoudness ...... 87 3.5 Discussion...... 87 3.5.1 The Frequency of BAL Variations ...... 89 3.5.2 Constraints Upon BAL Lifetimes ...... 89 3.5.3 Relation of BAL Disappearance and Emergence to General BAL Variability...... 91 3.5.4 A Random-Walk Model for The Evolution of BAL Troughs . . 92 3.5.5 Shielding Gas Changes as a Driver of Coordinated Multi-Trough Variability? ...... 96 3.6 SummaryandFutureWork ...... 97
Chapter 4. The Dependence of C iv Broad Absorption Line Properties on Accompany- ing Si iv and Al iii Absorption: Relating Quasar-Wind Ionization Levels, Kinematics, and Column Densities ...... 110 4.1 Introduction...... 110 4.2 Observations, Sample Selection, and Data Preparation ...... 113 4.3 Identification of BAL Troughs and Measurements ...... 115 4.3.1 Identification of BAL Troughs ...... 115 4.3.2 Measurements of BAL Troughs ...... 117 4.3.3 Comparisons with Mg ii, Fe ii, and P v ...... 127 4.4 Results...... 128 4.4.1 BAL-TroughProfiles ...... 128 4.4.2 C iv BAL-TroughStrengths ...... 128 4.4.3 C iv BAL-Trough Velocities ...... 132 vii
4.4.4 C iv BAL-Variation Profiles ...... 135 4.4.5 C iv BALEWVariability ...... 137 4.4.5.1 EW Variability as a Function of EW ...... 137 h i 4.4.5.2 BAL-Trough Samples with Matching EWs ...... 138 4.4.6 EW Variation Correlations ...... 141 4.4.6.1 EW Variation Correlations for the C ivS0 Sample . . 142 4.4.6.2 EW Variation Correlations for the C ivSA Sample . . 143 4.5 Summary of Results, Discussion, and Future Work ...... 146
Chapter 5. Summary ...... 152
Bibliography ...... 154 viii List of Tables
2.1 Sample of Quasars Showing Disappearing BAL Troughs ...... 39 2.2 Parameters of Disappearing BAL Troughs ...... 41 2.3 Observed Fractions and Lifetimes for Disappearing BAL Troughs ..... 42
3.1 Sample-Based Studies of BAL Quasar Variability ...... 101 3.2 C iv BALTroughs ...... 102 3.3 Si iv BALTroughs ...... 104 3.4 C iv BAL-Trough Variable Regions ...... 106 3.5 Si iv BAL-Trough Variable Regions ...... 107 3.6 Variability Comparison Between the Lowest Velocity BAL Trough and Ad- ditionalBALTroughs ...... 108 3.7 Spearman Rank Correlation Test Probabilities for Quasar Properties vs. BAL-TroughEWVariations...... 109
4.1 Main-SampleBALQuasars ...... 121 iv 4.2 C 00 Troughs ...... 122 iv 4.3 C S0 Troughs ...... 123 4.4 C ivSA Troughs...... 124 4.5 Average values of BAL-trough properties for C iv00,C ivS0,C ivSA, and all C iv BALtroughs...... 132 4.6 Average values of BAL-trough velocities for C iv00,C ivS0,C ivSA, and all C iv BALtroughs...... 135 4.7 Two-Sample Anderson-Darling Test Results for BAL-Trough Samples with Matching First-Epoch EWsa ...... 141 ix List of Figures
2.1 Comparison of the total rest-frame EWs of the C iv BALs...... 10 2.2 The best-fit continuum models for four example BOSS spectra of quasars withdisappearingBALtroughs...... 14 2.3 Multi-epoch spectra of quasars with disappearing BAL troughs obtained fromSDSS-I/IIandBOSS...... 17 2.4 Multi-epoch observations of the six quasars with disappearing BAL troughs thathave morethanoneSDSSorBOSSspectrum...... 23 2.5 Three-epoch observations of J093620.52+004649.2 that show the only re- ported example of a combined emergence and disappearance event for a BALtrough...... 24 2.6 BAL quasars that transformed to non-BAL quasars after the disappearance of the C iv BALtrough...... 28 2.7 Redshifts vs. absolute i-band magnitudes of main-sample quasars in this study and quasars with disappearing C iv BALtroughs...... 29 2.8 EW distributions for disappearing C iv BAL troughs, the other BAL troughs present in quasars that show one disappearing trough, and all 925 distinct C iv BALtroughsinthemainsample...... 31 25 iv 2.9 Distribution of fdeep for disappearing C BAL troughs and for all main sample...... 32 2.10 vmax, vmin, vc, and ∆v distributions for disappearing C iv BAL troughs and forallmainsample...... 33 2.11 The EWs at two epochs for all BAL troughs in the main sample and for other BAL troughs present in quasars that show one disappearing trough. 34 2.12 Fractional changes in EW for 12 additional non-disappearing C iv BAL troughs present in quasars that show one disappearing trough...... 35
3.1 Absolute i-band magnitude, Mi, vs. redshift for the main-sample quasars. 48 3.2 Comparison of the i-band apparent magnitude distributions of the main sample of this paper, the 2005 targets of the BOSS ancillary project on BAL quasars, and the BAL quasars identified in the SDSS DR10 quasar catalog...... 49 3.3 Example of our adopted BAL-trough definition illustrated using the three available spectra of the quasar SDSS J090944.05+363406.7...... 52 3.4 Distributions of the minimum sampled rest-frame timescale, ∆tmin, for 428 distinct C iv and 235 distinct Si iv BALtroughs...... 53 3.5 Cumulative fraction of BAL troughs with a given threshold ∆EW for C iv | | and Si iv troughs...... 56 x
3.6 Number of times a variable region is found at a particular velocity in C iv and Si iv BAL troughs for variations on timescales of more than 1 yr. . . 58 3.7 Percentage of variable regions with a given velocity width in C iv and Si iv BAL troughs for variations on timescales of more than 1 yr...... 60 3.8 Distribution of f∆v, and f∆v distribution as a function of BAL trough width ∆v...... 61 3.9 The number of variable regions found at a given vnrt...... 62 3.10 EW variation, ∆EW, vs. the minimum sampled rest-frame timescale, ∆tmin, for C iv and Si iv BALtroughs...... 63 3.11 Fractional EW variation, ∆EW/ EW , vs. the minimum sampled rest- iv h iiv frame timescale, ∆tmin, for C and Si BALtroughs...... 64 3.12 The fraction of C iv BAL troughs showing variability at more than 3σ significance as a function of timescale...... 65 3.13 Distributions of rate-of-change of EW , ∆EW /∆t, for C iv and Si iv BAL | | | | troughs both on short and long timescales...... 67 3.14 Distributions of BAL-trough EW variations, ∆EW, and fractional EW vari- ations, ∆EW/ EW , for C iv and Si iv for variations on timescales of more h i than1yr...... 68 3.15 EW variation, ∆EW, vs. average EW over the two relevant epochs, EW , h i for C iv and Si iv BAL troughs for three different timescales...... 71 3.16 Fractional EW variation, ∆EW/ EW , vs. average EW over the two rel- h i evant epochs, EW , for C iv and Si iv BAL troughs for three different h i timescales...... 72 3.17 C iv BAL-trough EW variation, ∆EW, as a function of BAL-trough width, ∆v, average depth of BAL troughs, dBAL, and centroid velocity, vcent. . . 74 3.18 C iv BAL-trough ∆EW/ EW as a function of EW and v for variations h i h i cent ontimescalesofmorethan1yr...... 75 3.19 Comparison of C iv and Si iv BAL-trough EWs for troughs having corre- spondingvelocities...... 76 3.20 Comparison of the variability of C iv vs. Si iv BAL troughs...... 77 3.21 Comparison of EW variations and fractional EW variations of distinct troughs for BAL quasars with multiple C iv troughs...... 80 3.22 Percentage of additional C iv BAL troughs at a given velocity offset. . . . 82 3.23 EW variability of C iv and Si iv BAL troughs as a function of quasar bolo- metric luminosity, LBol...... 84 3.24 EW variability of C iv and Si iv BAL troughs as a function of Eddington- normalized luminosity, LBol/LEdd...... 86 3.25 EW variation for C iv BAL troughs as a function of radio-loudness param- eter, R...... 88 3.26 Two-epoch spectra from SDSS and BOSS of quasars with BAL troughs that remain remarkably stable over multiple years in the rest frame. . . . 90 xi
3.27 Distribution of C iv BAL-trough EW variations, ∆EW, for variations on timescales of 2.0–2.5 yr and a binomial distribution calculated for a random- walk model with six steps and a step size of 1.65 A.˚ ...... 93 3.28 68.3%, 90%, and 99% confidence intervals for the two parameters of the random-walk model describing C iv BAL-trough EW variations, ∆EW, on timescalesof2.0–2.5yr...... 95 3.29 Comparison between EW variations for 428 C iv BAL troughs in our main sample and the best-fitting random-walk model...... 96
4.1 Two-epoch spectra of quasars with C iv00 BAL troughs ...... 118 4.2 Two-epoch spectra of quasars with C ivS0 BAL troughs ...... 119 iv 4.3 Two-epoch spectra of quasars with C SA BAL troughs ...... 120 4.4 The rest-frame ∆t distributions for all C iv troughs in our main sample . 126 iv iv iv 4.5 Composite BAL-trough profiles for C 00,C S0, and C SA BAL troughs 129 iv iv iv 4.6 Composite BAL-trough profiles for C 00,C S0, and C SA BAL troughs as a function of normalized C iv troughwidth...... 130 4.7 Average EW distributions for C iv00,C ivS0, and C ivSA BAL troughs . . 131 4.8 Average depth and velocity width distributions for C iv00,C ivS0, and C ivSA BALtroughs ...... 133 4.9 Minimum velocity, maximum velocity, and centroid velocity distributions for C iv00,C ivS0, and C ivSA BALtroughs ...... 134 4.10 Depth variation profiles for C iv00,C ivS0, and C ivSA BAL troughs . . . . 136 iv iv iv 4.11 ∆EW vs. EW for C 00,C S0, and C SA BAL troughs ...... 139 h i iv iv iv 4.12 ∆EW/ EW vs. EW for C 00,C S0, and C SA BAL troughs . . . . 140 h i h i iv iv iv iv 4.13 ∆EW distributions for C 00 vs. C S0 troughs and for C S0 vs. C SA troughs, where the samples show BAL troughs with similar EWs...... 142 iv iv iv 4.14 ∆EW as a function of EW1 for C 00,C S0, and C SA BAL troughs . 143 4.15 The ∆EW and ∆EW/ EW correlations between C iv BAL troughs and h i S0 the corresponding Si iv BALtroughs...... 144 4.16 The ∆EW and ∆EW/ EW correlations for C iv vs. Si iv troughs and h i SA C ivSA vs. Al iii troughs ...... 145 4.17 Density and poloidal velocity maps of the disk-wind model...... 149 xii Acknowledgments
I would like to express my sincere thanks to my spouse Hasan Ak and my parents Safiye and Nazmi Filiz for their invaluable support and patience during my long journey. I also would like to thank my son Omer Yagiz Ak for bringing the joy to my life and giving me the motivation to pursue my degree. I am very grateful to my PhD advisor Niel Brandt for his guidance, time and effort throughout my study. He has been providing assistance in every step of my study and his high expectations help me develop and improve my academic skills. Without his guidance and persistent help this dissertation would not have been possible. I thank all of my dissertation committee members for their constructive comments and useful advices. I am particularly thankful to Don Schneider and Mike Eracleous for their invaluable contributions to my studies and to my academic development. I also like to thank my collaborators Pat Hall, Jonathan Trump, Scott Anderson, Rob Gibson, Fred Hamman, Brit Lundgren, Adam Myers, Isabelle Paris, Patrick Petitjean, Nic Ross, Yue Shen and Don York for their invaluable contribution to my studies. I am grateful to have Ana Matkovic, Rohit Deshpande, and Suvrath Mahadevan as my mentors and friends, who, directly or indirectly, provided me help, encouragement, and support since the first day we have met. I take this opportunity to record my sincere thanks to all faculty, staff and students of the Astronomy Department for their help and encouragement. I also place on record, my sense of gratitude to one and all friends I have met during these four years in Penn State. In particular, I would like to thank my friends; Prakash Arumugasamy, Lea and Alex Hagen, Jackson Norris, Ryan Terrien, and Sharon Wang. I was so fortunate to be a member of SDSS and in particular BOSS collaboration. Collaboration meetings and workshops provide me a unique opportunity to meet important names in my research field and encourage me to carry my study further. 1
Chapter 1
Introduction
In this thesis, I utilize spectroscopic observations from the Sloan Digital Sky Survey (SDSS) and the Sloan Digital Sky Survey-III (SDSS-III) Baryon Oscillation Spectroscopic Survey (BOSS) to study broad absorption line (BAL) variability on multi-year timescales. Repeated observations of BAL quasars in SDSS and BOSS surveys provides a unique oppor- tunity to investigate variability of quasar outflows and their relation to quasar properties.
1.1 Active Galactic Nuclei
A small fraction of galaxies exhibit non-thermal radiation over almost all wave- lengths of the electromagnetic spectrum. This typical energy distribution distinguishes them from normal galaxies that are simply collections of stars, dust and gas. These un- usual galaxies are known as active galaxies, in reference to the active galactic nuclei (AGN) in their central region. The tremendous luminosities of AGNs, the compactness of their central regions, and their non-thermal broadband energy distributions indicate that the powering mechanism is likely to be accretion rather than nuclear reactions. However, we do not have a direct observational proof indicating accreting material on the central supermassive black hole (SMBH). Owing to the differences in observed properties, AGNs are classified into subgroups: low luminosity AGNs (e.g., Seyfert galaxies), the most luminous AGNs (e.g., quasars), strong radio emitters (e.g., radio galaxies), and very strong and rapid optical variables (e.g., BL Lac objects). However, further observations suggest that some of these subgroups are in fact part of a continuous sequence with no discrete physical difference. For example, Seyfert galaxies with high luminosities are indistinguishable from quasars and a small minority of quasars (10-15%) are also strong radio emitters. These similarities between the subgroups encourage unification models in which the classification of an AGN depends strongly on the observer’s location and viewing angle.
1.2 Broad Absorption Line Quasars
In addition to the strong emission lines, 30% of quasars present broadened and ≈ almost always blueshifted absorption lines, relative to related emission lines, in their UV and X-ray spectra. The relatively high (i.e. & 0.1c) velocities and blueshifted nature of these absorption lines indicates that they are produced by an absorption system in which 2 absorbing material is carried out by outflowing winds along the line of sight. Absorption systems in quasars can be associated with the active nucleus and are thus classified as “intrinsic” while the ones originate from clouds external to the central environment are classified as “intervening”. Broad absorption lines (BAL) are the most prominent imprints of intrinsic absorp- tion systems seen in quasar spectra. Absorbing material launched from the central region of an AGN and carried away by outflowing winds is particularly consequential for two main reasons. First, ongoing accretion onto the SMBH requires expulsion of the angular mo- mentum from the system. An outflowing wind carries some amount of material from the central environment, along with angular momentum, maintaining the accretion. Second, outflowing winds of an AGN may be responsible agents of feedback into host galaxies; the material carried along with these winds may regulate star formation in the host galaxy and shape galaxy evolution. If outflows are driving agents of continuing accretion and star formation in the host galaxies, why are they observed only in a relatively small fraction of the quasars? It is possible that all the quasars have outflows but most are not in our line-of-sight at a given time. The basic approach to investigate this idea is to assess the similarities of physical properties for BAL and non-BAL quasars. The first large sample of quasars is analyzed by Weymann et al. (1991). Weymann et al. (1991) define a BAL quasar to have at least one BAL trough which is at least 2000 km s−1 wide at 10% under the continuum level. Based on this definition, only up to 10–20% of quasars are identified as BAL quasars (Weymann ≈ 1991, Hewett & Foltz 2003, Reichard et al. 2003, Trump et al. 2006, Gibson et al. 2009). Considering optical/UV selection effects, the percentage rises to 17–23% (Hewett & Foltz ≈ 2003, Knigge et al. 2008, Gibson et al. 2009). The fraction of BAL quasars is strongly dependent upon the definition of a BAL trough. Using a less restrictive criteria, Trump et al. (2006) show that the fraction of BAL quasars could rise up to 20–40%. A comparison between composite spectra of BAL and non-BAL quasars reveals the most significant difference between these two groups of quasars; BAL quasar spectra tend to be more dust obscured (e.g., Trump et al. 2006; Gibson et al. 2009). Consequently, multi- wavelength studies showed that BAL quasars are generally relatively X-ray weak compared to non-BAL quasars (e.g. Gallagher et al. 2006). These findings lead to improved models of BAL quasars that satisfy observational constrains and hence a better understanding of BAL quasar environment. The most compelling model of BAL quasars indicates that outflows are radiatively driven from the disk surrounding the central black hole, and a shielding gas prevents BAL outflows from being over-ionized by highly energetic emission generated close to the central region of a quasar (Murray et al. 1995; Proga et al. 2000). Moreover, the shielding gas is found to be variable in wind simulations (Sim et al. 2010), indicating a connection to the variable nature of BAL troughs. Multi-epoch observations of BAL quasars revealed that variability is the most promi- nent characteristic of BAL troughs; the time variable absorption strength is commonly 3 observed in UV spectra (Barlow 1993; Lundgren et al. 2007; Gibson et al. 2008, 2009, 2010; Capellupo et al. 2011, 2012). Previous studies also present a few examples for dis- appearance (Junkkarinen et al. 2001; Lundgren et al. 2007; Hall et al. 2011; Vivek et al. 2012) and emergence (Ma 2002; Lundgren et al. 2007; Hamann et al. 2008; Leighly et al. 2009; Krongold et al. 2010; Capellupo et al. 2012; Vivek et al. 2012) of absorption troughs in quasar spectra. However, these studies presented only a few examples of BAL quasars in spectral observations spanning limited timescales. Systematic observations of a large BAL-quasar sample on multi-year timescales will help to determine physical parameters of the absorption systems, dynamics of the quasar outflows and lifetimes of the BAL troughs.
1.3 A Large-Scale Survey of Quasar BAL Variability with SDSS BOSS
Modern sky surveys are capable of delivering multi-epoch observations for enormous samples of quasars. Sloan Digital Sky Survey (SDSS) is one of the most recent example of such large surveys (York et al. 2000). SDSS is a large multi-filter imaging and spectroscopic survey using a dedicated 2.5-m optical telescope (Gunn et al. 1998, 2006) at Apache Point Observatory in New Mexico. During its first phase of operations, 2000–2005, the SDSS imaged more than 8,000 square degrees of the sky in five optical bandpasses, and it obtained spectra of galaxies and quasars. In 2005 the survey entered a new phase, the SDSS-II, expanding its samples to over 800,000 galaxies and 100,000 quasars. The Baryon Oscillation Spectroscopic Survey (BOSS), part of the third phase of SDSS operations, is designed to obtain spectra for 1.5 million luminous galaxies and 160,000 quasars (e.g., Eisenstein et al. 2011; Anderson et al. 2012; Ross et al. 2012; Dawson et al. 2013). The main aim of BOSS is to obtain precision measurements of the cosmic distance scale using baryon acoustic oscillations. The BOSS survey started operating in mid-2008 and it is planned to continue observations until the end of 2014. The BOSS quasar sur- vey focus upon quasars with z > 2.1 and that provides an outstanding opportunity for discovering intrinsic Si iv and C iv absorptions in quasars (Smee et al. 2013). In addition to its main aim, the BOSS hosts several ancillary projects. One of these projects aims to obtain spectra of about 2000 relatively bright BAL quasars discovered in the SDSS-I and SDSS-II surveys. These repeat observations are designed for a high-quality study of BAL variability over multi-year timescales in the rest frame. The target sample of the project is about two orders of magnitude larger than samples of previous multi-epoch BAL variability studies.
1.4 Outline of the Thesis
In this thesis, I investigate characteristics of BAL troughs and their variations on multi-year timescales using spectroscopic observations from SDSS-I/II and BOSS. Each chapter is a self-contained work with comprehensive introductions and motivations. 4
In Chapter 2, I present 21 examples of C iv BAL trough disappearance in 19 quasars selected from systematic multi-epoch observations of 582 bright BAL quasars x(1.9 Several chapters of this dissertation have been published. The references to the related articles are listed below: 1. Broad Absorption Line Disappearance on Multi-Year Timescales in a Large Quasar Sample; Filiz Ak, N., Brandt, W. N., Hall, P. B., et al. 2012, ApJ, 757, 114 2. Broad Absorption Line Variability on Multi-Year Timescales in a Large Quasar Sam- ple; Filiz Ak, N., Brandt, W. N., Hall, P. B., et al. 2013, ApJ, 777, 168 3. Broad Absorption Line Variability as a Function of Outflow Column Density; Filiz Ak, N., Brandt, W. N., Hall, P. B., et al. 2014, ApJ, in prep. 6 Chapter 2 Broad Absorption Line Disappearance on Multi-Year Timescales in a Large Quasar Sample 2.1 Introduction Intrinsic absorption lines in quasar spectra are often produced by outflowing winds along the line of sight that are launched from the accretion disk or other structures around the central supermassive black hole (SMBH; e.g., Murray et al. 1995; Proga et al. 2002). Such absorption lines, shaped by the geometry and kinematic structure of the outflows, appear in quasar spectra as broad absorption lines (BALs; ∆v 2000 km s−1), mini-BALs ≥ (2000 ∆v 500 km s−1), or intrinsic narrow absorption lines (NALs; ∆v 500 km s−1); ≥ ≥ ≤ e.g., see Hamann & Sabra (2004) and Ganguly & Brotherton (2008). Traditionally defined BAL troughs are observed in 15% of quasars; these troughs are sufficiently strong that ≈ the depths of the features lie at least 10% under the continuum in the, e.g., Si iv λ 1400, C iv λ 1549, Al iii λ 1857, or Mg ii λ 2799 transitions with blueshifted velocities up to 0.1c ≈ (e.g., Weymann et al. 1991; Gibson et al. 2009, hereafter G09; Allen et al. 2011; and refer- ences therein). Quasars that present BAL troughs in their spectra are often classified into three subtypes depending on the presence of absorption lines in specified transitions: (1) High-ionization BAL quasars show absorption lines in high-ionization transitions including Si iv, and C iv. (2) Low-ionization BAL quasars possess Mg ii and/or Al iii absorption lines, in addition to the high-ionization transitions. (3) Iron low-ionization BAL quasars show additional absorption from excited states of Fe ii and Fe iii (e.g., Hall et al. 2002, and references therein). The winds revealed by intrinsic quasar absorption lines are of importance for two main reasons. First, the observed frequency of quasar absorption lines indicates that these winds are a significant component of the nuclear environment. Indeed, disk accretion onto the SMBH may require significant mass ejection for expulsion of angular momentum from the system (e.g., Blandford & Payne 1982; Crenshaw et al. 2003). Second, quasar winds may be agents of feedback into massive galaxies, regulating star formation and further SMBH accretion via the removal of cold gas (e.g., Springel et al. 2005; King 2010). BAL troughs, the strongest observed signatures of quasar winds, often vary in equiv- alent width (EW) and/or shape over rest-frame timescales of months to years (e.g., Barlow 1993; Lundgren et al. 2007; Gibson et al. 2008, 2010; Capellupo et al. 2011, 2012). Recent statistical studies of BAL variability have shown that the fractional EW change increases 7 with rest-frame timescale over the range 0.05–5 yr. Such variations could, in principle, be driven by changes in covering factor, velocity structure, or ionization level. Of these pos- sibilities, the generally favored driver for most BAL variations is changes in the covering factor of outflow stream lines that partially block the continuum emission (e.g., Hamann 1998; Arav et al. 1999b). These covering-factor changes could ultimately be caused by ro- tation of a non-axisymmetric outflow that is loosely anchored to the accretion disk, or they could arise from large-scale changes in wind structure (e.g., Proga et al. 2002). Changes in ionization level are generally disfavored as a primary driver, since BAL troughs are often highly saturated and thus should be only weakly responsive to ionization-level changes. Furthermore, BAL trough variations generally do not appear to be correlated with varia- tions of the observable continuum (typically) longward of Lyα (e.g., Gibson et al. 2008), although the observable continuum is not the ionizing continuum for the BAL gas (i.e., it is possible that the two may vary differently). Strong variations in BAL EWs observed over multi-year timescales in the rest frame suggest that BAL disappearance and BAL emergence may be significant effects on such timescales (e.g., Gibson et al. 2008; Hall et al. 2011). However, largely owing to practi- cal difficulties in observing large samples of BAL quasars on multi-year timescales (often corresponding to 10–20 yr in the observed frame), only a small number of BAL disappear- ance (Junkkarinen et al. 2001; Lundgren et al. 2007; Hall et al. 2011; Vivek et al. 2012) and emergence (Ma 2002; Lundgren et al. 2007; Hamann et al. 2008; Leighly et al. 2009; Krongold et al. 2010; Capellupo et al. 2012; Vivek et al. 2012) events have been discov- ered. Specifically, considering the BAL disappearance phenomenon of most relevance to this paper, we are only aware of four reported cases. In the first, a Mg ii BAL in the spectrum of the binary quasar LBQS 0103–2753 essentially disappeared over 6.0 rest- ≤ frame years, converting this object from a low-ionization BAL quasar to a high-ionization BAL quasar (Junkkarinen et al. 2001). In the second, the highest velocity C iv trough in the spectrum of J075010.17+304032.3 nearly disappeared in 0.3 rest-frame years, which ≤ is the only known example of C iv disappearance (Lundgren et al. 2007). In the third, the Fe ii troughs in the spectrum of FBQS J1408+3054 disappeared over 5.1 rest-frame ≤ years, converting this object from an iron low-ionization BAL quasar to a low-ionization BAL quasar (Hall et al. 2011). In the fourth, a Mg ii BAL trough in the spectrum of SDSS J133356.02+001229.1 disappeared in 3.7 rest-frame years and another Mg ii BAL ≤ trough — that emerged at higher velocity — nearly disappeared in 5.7 rest-frame years ≤ (Vivek et al. 2012). These disappearance examples did not involve transformations from BAL quasars to non-BAL quasar status because troughs from other ions, or additional C iv troughs, remained. We have compiled a sample of 19 quasars with disappearing BAL troughs from a well-defined, large sample of BAL quasars whose spectra were observed over 1.1–3.9 yr in the rest frame. We aim to define the basic statistical characteristics of the BAL disappear- ance phenomenon. In this study, we will discuss the disappearance of BAL troughs as well as the transformation of BAL quasars to non-BAL quasars. These two phenomena can 8 be distinct given that some BAL quasars have multiple troughs. A quasar that possesses more than one BAL trough could have only one of them disappear without transforming to a non-BAL quasar. We only consider BAL quasars without any remaining BAL troughs at a later epoch (in any transition) to have transformed from a BAL quasar to a non- BAL quasar. The observations and the selection of the main sample used in this study are discussed in Section 2.2. Identification of disappearing BAL troughs is explained in Section 2.3. Statistical results and comparisons are presented in Section 2.4. A summary of the main results and some future prospects are given in Section 2.5. −1 −1 Throughout this work we use a cosmology in which H0 =70 km s Mpc ,ΩM = 0.3, and ΩΛ = 0.7. All timescales are in the rest frame unless otherwise mentioned. 2.2 Observations and Sample Selection 2.2.1 Observations The Baryon Oscillation Spectroscopic Survey (BOSS), part of the Sloan Digital Sky Survey-III (SDSS-III; Eisenstein et al. 2011), is a five-year program (2009–2014) that is using the 2.5-m SDSS telescope (Gunn et al. 2006) at Apache Point Observatory to obtain spectra for 1.5 million luminous galaxies as well as 160, 000 quasars at z > 2.2 ≈ ≈ 2 (Anderson et al. 2012; Ross et al. 2012). These targets are selected from 10, 000 deg ≈ of sky at high Galactic latitude. While the primary goal of BOSS is to obtain precision measurements of the cosmic distance scale using baryon acoustic oscillations, the resulting spectra (covering 3600–10000 A˚ at a resolution of 1300–3000; see Eisenstein et al. 2011) are valuable for a wide range of investigations, including studies of quasar physics. In addition to the main BOSS galaxy and quasar surveys, a number of smaller ancillary BOSS projects are being executed. One of these ancillary projects, relevant to this paper, focuses on studying BAL variability on multi-year timescales in the rest frame. The main goal of this project is to move from small-sample and single-object studies of multi-year BAL variability to setting large-sample statistical constraints that can ultimately be compared with models of quasar winds. The basic approach is to obtain BOSS spectroscopy of BAL quasars already observed from 2000–2008 by SDSS-I/II. Most of the targets for this project are drawn from the G09 catalog of 5039 BAL quasars in the SDSS Data Release (DR) 5 quasar catalog (Schneider et al. 2007). Specifically, we have selected the 2005 BAL quasars from this catalog that are optically bright (i< 19.3); have high signal-to-noise ratio SDSS-I/II spectra (SN1700 > 6 as defined by G09, when SN1700 measurements are available); have full spectral coverage of their C iv, Si iv, Mg ii, or Al iii BAL regions (implemented via the redshift cuts described in Section 4 of G09); and have −1 moderately strong to strong BAL troughs (balnicity indices of BI0 > 100 km s as defined by G09). This uniformly defined sample is 100 times larger than current data sets that ≈ have been used to study BAL variability on multi-year timescales in the rest-frame (e.g., Gibson et al. 2008, 2010; Capellupo et al. 2011, 2012). We are also targeting 102 additional 9 BAL quasars selected to include unusual BAL quasars (e.g., Hall et al. 2002); objects with multiple SDSS-I/II observations; and objects with historical coverage by the Large Bright Quasar Survey (e.g., Hewett et al. 1995) or the First Bright Quasar Survey (e.g., White et al. 2000). Observations for our ancillary project began at the same time as the primary BOSS observations, and in this paper we will utilize spectra taken after the end of hardware commissioning (MJD 55176; see Ross et al. 2012) until MJD 55811 (i.e., 2009 December 11 until 2011 September 7). Within this date range, 692 of our 2005 primary BAL quasars were targeted, and observations for this program will continue throughout the BOSS project. 2.2.2 Sample Selection For the purpose of the BAL disappearance studies in this paper, we have selected a subset of the 2005 targeted BAL quasars with both SDSS-I/II and BOSS observations that satisfy our selection criteria given in Section 2.2.1 — all 102 unusual and other BAL quasars are excluded. We will focus on C iv BALs, as this is the most commonly studied strong BAL transition and has limited blending and confusion with other nearby transitions. Following Section 4 of G09, we therefore consider only those BAL quasars in the redshift range z = 1.68–4.93 where the SDSS-I/II spectra provide full coverage of the C iv region. As in past work (e.g., Lundgren et al. 2007), we only consider C iv BAL troughs that are significantly detached from the C iv emission line; this selection minimizes the chance of confusion between BAL variability and emission-line variability. To implement this requirement formally, we consider only those C iv BAL trough regions lying in the velocity range of 30000 v < 3000 km s−1 (v is the maximum observed velocity for the − ≤ max − max BAL trough region and is defined fully in Section 2.3.2). We show in Section 2.4.3 that this requirement should not significantly bias our statistical results. The v 30000 km s−1 max ≥− requirement minimizes confusion between C iv BALs and the Si iv emission line. The selections above result in 582 BAL quasars which we define as our “main sam- ple”. By construction, all of these quasars have at least one SDSS-I/II spectrum and one BOSS spectrum. A significant minority have additional SDSS-I/II or BOSS spectra. In total, we have 1396 spectra for the 582 main-sample BAL quasars. The rest-frame time intervals between spectral observations range from 0.26 days to 3.86 yr, although in this paper our emphasis will be on > 1 yr timescales. Figure 2.1 shows the average total EW versus the maximum sampled rest-frame timescale, ∆tmax. The EW values of all C iv BAL troughs in a given spectrum are summed and averaged over two epochs (the first SDSS and last BOSS spectrum). The maximum timescales for our sample range between 1.10 and 3.88 yr, with a mean of 2.52 yr and a median of 2.50 yr. For comparison, we have also plotted several other samples of BAL quasars whose BAL variability properties have been previously investigated (Barlow 1993; Lundgren et al. 2007; Gibson et al. 2008; Capellupo et al. 2011). Note the relatively large size of our sample compared to past work as well as the fairly long timescales being probed. 10 12000 Lundgren et al. (2007) 10000 Barlow (1993) This study ) Gibson et al. (2008) −1 8000 Capellupo et al. (2011) 6000 4000 2000 0 0.1 0.5 1 5 Rest−Frame ∆ t (years) max Fig. 2.1: Comparison of the total rest-frame EWs (summed over all BAL troughs and averaged over two epochs) of the C iv BALs for the sources in Barlow (1993; green stars), Lundgren et al. (2007; blue triangles), Gibson et al. (2008; red diamonds), Capellupo et al. (2011; cyan squares), and our study of the 582 BAL quasars in the main sample (purple circles). The x-axis shows the maximum sampled timescale in the rest frame. Dark purple circles denote the quasars with disappearing troughs. The EW is shown in units of kms−1; 1 A˚ is about 200 km s−1 in the C iv region. 11 We have cross correlated our 582 main-sample BAL quasars with the Shen et al. (2011) catalog to check for radio emission detected in Very Large Array (VLA) Faint Images of the Radio Sky at Twenty-Centimeters (FIRST; Becker et al. 1995) observations. Shen et al. (2011) list the radio-loudness parameter defined as R = f /f , where f is 6cm 2500A˚ 6cm the radio flux density at rest-frame 6 cm and f is the optical flux density at rest-frame 2500A˚ 2500 A.˚ We also have examined the NRAO VLA Sky Survey (NVSS; Condon et al. 1998) for three quasars that are not located in the FIRST footprint. Radio emission is detected from 55 of the main-sample quasars; seven of them are radio loud with R 100. ≥ 2.3 Identification of Disappearing BALs 2.3.1 Continuum Fit and Normalization To study variations in BAL characteristics, we investigate multi-epoch spectra that are normalized by an estimated continuum model. To determine proper continuum levels, we first corrected the main-sample spectra for Galactic extinction using a Milky Way extinction model (Cardelli et al. 1989) for RV = 3.1. The AV values were taken from the Schlegel et al. (1998). We then translated observed wavelengths to the rest frame using redshifts from Hewett & Wild (2010). The data-reduction pipelines for SDSS I/II (DR7) and BOSS (v5 4 45)1 remove night-sky lines from the data, but occasionally there are significant residuals near promi- nent night-sky lines. The data-processing algorithm flags the afflicted pixels in pixel-mask columns. We examine the “BRIGHTSKY” mask column and remove the flagged pixels from each spectrum. As in G09, we select the following six relatively line-free (RLF) windows, if they have spectral coverage, to fit a continuum model to each spectrum: 1250–1350 A,˚ 1700–1800 A,˚ 1950–2200 A,˚ 2650–2710 A,˚ 2950–3700 A,˚ and 3950–4050 A.˚ We select these RLF windows to be free from strong emission and absorption in general and to represent the underlying continuum both blueward and redward of the C iv and Si iv BAL regions; the use of the RLF windows spanning a fairly broad range of wavelengths provides useful leverage for constraining the continuum shape. We avoid the use of any window at shorter wavelengths due to heavily absorbed regions, such as H i absorption in the Lyα forest. Previous studies of BAL variability have used a variety of models to define the un- derlying continuum, such as power-law fits (e.g., Barlow 1993; Capellupo et al. 2011, 2012), reddened power-law fits (e.g., Gibson et al. 2008, 2009), and polynomial fits (e.g., Lundgren et al. 2007). As in G09, we prefer to use a reddened power-law that reconstructs the under- lying continuum well at all covered wavelengths with a small number (three) of parameters. Hopkins et al. (2004) show that the intrinsic dust reddening in quasar spectra is dominated by Small-Magellanic-Cloud-like (SMC-like) reddening. We reconstruct the continuum with 1The BOSS pipeline is described in Aihara et al. (2011). 12 a power-law model that is intrinsically reddened using the SMC-like reddening model from Pei (1992). The three continuum-model parameters are thus the power-law normalization, the power-law spectral index, and the intrinsic-absorption coefficient. The continuum-model parameters for each spectrum are obtained from a non-linear least-squares fit with an iterative “sigma-clipping” algorithm. The sigma-clipping filtering iteratively excludes the data points that deviate by more than 3σ from the previous fit. Absorption or emission features (e.g., intervening absorption lines) that are occasionally present in the RLF windows are thereby filtered. Typically 1–8% of the RLF-window spectral pixels are excluded in a given spectrum. We calculate the continuum-model pa- rameters with 68.3% confidence bounds (given that the average line-flux sensitivity in SDSS spectra matches the 1σ noise; Bolton et al. 2004). We calculate the confidence bounds us- ing numerical ∆χ2 confidence region estimation for three parameters of interest; see, e.g., Section 15.6.5 of Press et al. (1992). We propagate the derived continuum uncertainties through to continuum-normalized flux densities and EW measurements (see Section 2.3.2 and Table 2.2). We do not assign physical meaning to the resulting continuum-model parameters, because (1) degeneracy in model parameters allows multiple sets of parameters to produce nearly the same continuum, and (2) flux levels in blue regions (3500–5500 A˚ in the observed frame) of BOSS spectra are not yet absolutely or relatively calibrated to better than 10% ± overall, with greater uncertainties at shorter wavelengths owing to remaining instrumental effects (e.g., Margala & Kirkby 2011). For further discussion of difficulties in the physical interpretation of continuum-model parameters, see Section 2.1 of Gibson et al. (2008). Figure 2.2 shows the best continuum-model fits for some examples of BOSS spectra. The RLF windows for each spectrum have miscellaneous line-like features that are itera- tively excluded by our sigma-clipping filtering. Visual inspection shows that our continuum- reconstruction algorithm produces appropriate continuum levels. Comparison of local con- tinua blueward and redward of C iv BAL troughs (not in the RLF windows) shows good agreement between the data and the fitted continuum. Furthermore, there is good agree- ment between two epochs that are fitted independently. The reddened power-law model includes extinction that changes only slowly with wavelength and cannot produce false BALs given these local continuum constraints. In addition to the statistical errors that we have quantified above, we also consid- ered possible systematic errors in the continuum fitting. For example, we considered the systematic errors arising from our sigma-clipping filtering technique. As a test of the ro- bustness of this approach, we repeated the filtering for 2σ and 4σ clipping and obtained essentially the same continuum fits as with 3σ clipping. We further estimated systematic errors in continuum models with one-sided sigma-clipping filtering. Sigma clipping of only positive deviations and of only negative deviations give lower and upper bounds on the continuum, respectively. We found that the estimated systematic errors with this partic- ular method are less than the statistical errors. Another possible systematic error could arise when the spectral region blueward of the Si iv BAL region lies near the edge of the 13 observed spectrum where there is significant noise. However, a visual comparison of the continuum-fitting results for such objects does not show any significant difference from the others. 2.3.2 Measurements of BAL Properties As is a common practice for BAL studies (e.g., Lundgren et al. 2007; Gibson et al. 2008, 2009), we smoothed each spectrum with a Savitzky-Golay (SG; Savitzky & Golay 1964) algorithm before investigating the BAL troughs. The SG parameters were selected to perform local linear regression on three consecutive data points that preserve the trends of slow variations and smooth the fluctuations originating from noise. We used the smoothed spectra only to facilitate C iv BAL detection. We searched each smoothed spectrum for absorption troughs at a level of 10% ≥ under the continuum as given in the traditional definition of BAL and mini-BAL troughs. Velocities corresponding to the shortest and the longest wavelengths for a given trough, vmax and vmin respectively, determine each trough’s width. We sorted the features into mini-BAL troughs (500–2000 km s−1 wide) and BAL troughs ( 2000 km s−1 wide); see ≥ Section 2.1 for further discussion. We detected a total of 925 distinct C iv BAL troughs in observations of 582 main- sample quasars. As described in 2.2, we consider only troughs lying in the velocity range of 30000 v < 3000 km s−1 to minimize confusion between BAL variability and − ≤ max − emission-line variability. For a small fraction of the distinct C iv BAL troughs (15%), the trough region extends beyond the given velocity limits, mainly at the low velocity boundary. For such BAL troughs we assign the relevant cut-off velocity as the boundary of the BAL trough, and we consider the portion of the trough that lies within the given limits for further calculations. A visual comparison of the detected BAL troughs and the positions of these troughs with those in G09 shows good general agreement. We calculate the C iv balnicity index (BI′) of each spectrum. We use a similar expression to the traditional BI definition given by Weymann et al. (1991). Our expression is −30000 f(v) BI′ 1 Cdv (2.1) ≡ Z−3000 − 0.9 In this definition, BI′ is expressed in units of km s−1 where f(v) is normalized flux density as a function of velocity, and C is a constant which is equal to 1.0 only where a trough is wider than 2000 km s−1, and is 0.0 otherwise. The only difference from the traditional balnicity definition is the limiting velocities of the C iv BAL region (i.e., from 3000 to − 30000 km s−1 instead of from 3000 to 25000 km s−1).2 − − − We calculate the rest-frame equivalent width (EW) for each BAL trough in units of both A˚ and kms−1; note that 1 A˚ corresponds to 200 km s−1 in the C iv absorption ≈ 2Negative signs indicate that the BAL trough is blueshifted with respect to the systemic velocity. 14 18 16 α Ly N V Si IV C IV C III J004022.40+005939.6 Mg II λ 14 z=2.565 12 10 8 Flux Density F 6 4 2 0 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 Rest Wavelength 20 C IV Si IV C III Mg II J094806.58+045811.7 λ 15 z=1.737 10 Flux Density F 5 0 1500 2000 2500 3000 3500 Rest Wavelength 70 60 α Ly N V Si IV C IV C III J132216.24+052446.3 Mg II λ 50 z=2.05 40 30 Flux Density F 20 10 0 1500 2000 2500 3000 Rest Wavelength 60 α Mg II N V C IV Si IV C III Ly 50 J155119.14+304019.8 λ z=2.410 40 30 Flux Density F 20 10 0 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 Rest Wavelength Fig. 2.2: The best-fit continuum models for four example BOSS spectra of quasars with disappearing BAL troughs. The observed flux density (in units of 10−17 erg cm−2 s−1 A˚−1) and the rest-frame wavelength (in A)˚ of each object (black) is fitted with the continuum model (solid red) in the plotted RLF windows (blue horizontal lines). Discrete spectral features within the RLF windows were removed via the iterative continuum fitting process. 15 region in the rest frame. EW and EW uncertainties are calculated from unsmoothed data using Equations 1 and 2 of Kaspi et al. (2002). In addition, following Gallagher et al. 25 (2006), we define the fdeep parameter to represent the fraction of data points that lie at least 25% under the continuum level in each BAL trough. As discussed in Gibson et al. (2008), BI′ and EW measurements alone are not satis- factory for variability studies. For the purpose of better understanding spectral variations, we compared observations of a given quasar at two different epochs that were obtained with a time interval ∆t. We define ∆t as the rest-frame interval between the last obser- vation showing a BAL trough (at time t1) and the first observation where the trough has disappeared (at time t2). Note that ∆t can be different from ∆tmax in Section 2.2.2 for the quasars that are observed in more than two epochs; ∆t represents an upper limit for the disappearance time of a BAL trough with the EW value seen in the last epoch in which the trough was present. For our sample, the range of ∆t is 1.10–3.88 yr with a mean of 2.47 yr and a median of 2.45 yr. The BOSS spectra are provided at the same observed wavelengths as the SDSS spectra (except for the additional BOSS wavelength coverage), therefore we can compare spectra at identical wavelengths. Proper comparison requires consideration of differences in signal-to-noise ratio (S/N) incorporating standard deviations (σ). Therefore, each (un- smoothed) pixel in the spectra, observed at t1 and t2, is compared using the following calculation of the quantity Nσ: f2 f1 Nσ(λ)= − (2.2) 2 2 σ2 + σ1 q where f1 and f2 are the normalized flux densities and σ1 and σ2 are the flux density standard deviations at wavelength λ. Both σ1 and σ2 include uncertainties from the continuum model (see Section 2.3.1), in addition to the uncertainties from the observations. To summarize Nσ is a measurement of the deviation between two observations for each pixel in units of σ (see Figure 2.3). 2.3.3 Selection of Disappearing BALs BAL troughs are often complex and can be difficult to quantify with simple and standard rules. Nevertheless, we define objective criteria to search all 582 main-sample BAL quasars for disappearing C iv BAL troughs. These criteria were based on the comparison between two normalized spectra, S1 and S2, observed at epochs t1 and t2. The utilized criteria were the following: The spectral region in S corresponding to the trough seen in S (lying between v • 2 1 min and vmax as defined in Section 2.3.2, and corresponding to the gray shaded regions in Figure 2.3) should be free from any BALs or mini-BALs. However, we do allow 16 the trough region in S2 to contain residual NALs. Our upper limit for any remaining iv C absorption in S2 will be discussed below. A two-sample χ2 test (see Section 4.4 of Bevington & Robinson 2003) comparing the • −8 data points in the trough region between S1 and S2 gives a probability of Pχ2 < 10 of there being no change. This criterion establishes that there has been a highly significant change in the trough region between S1 and S2; variations with less- significant values of Pχ2 may be real in some cases but can be arguable allowing for both statistical and systematic errors (see Section 2.3.1), as well as the general complexity of BAL troughs. The chosen probability threshold has been selected based upon visual inspection. Following the stated criteria, we identify 21 examples of disappearing C iv BAL troughs in 19 distinct quasars; these 19 quasars have ∆t values of 2.0–3.3 yr. Figure 2.3 compares normalized SDSS and BOSS spectra in the C iv and Si iv regions for quasars with disappearing BAL troughs. N σ values are also plotted for each wavelength bin. We have defined BAL disappearance following the formal definitions of BALs and mini-BALs given in Section 2.3.2. However, in the BOSS spectra of some objects there is weak residual absorption that fails our formal definition (see Section 2.3.2 and espe- cially Equation 1) for being a BAL or mini-BAL. We define 11 of the 21 examples of disappearing C iv BAL troughs as a “pristine” sample that has no significant remaining absorption (NAL regions excluded). These pristine sample objects are noted with stars in Figure 2.3. Table 2.1 lists all observation epochs for the quasars with disappearing BAL troughs in addition to redshifts from Hewett & Wild (2010), SDSS i-band magnitudes from Schneider et al. (2007), and absolute i-band magnitudes from Shen et al. (2011). The listed Plate-MJD-Fiber numbers are unique for each spectrum, and MJD 55176 indicates ≥ BOSS spectra. Table 2.2 lists the parameters of disappearing BAL troughs measured in 25 S1 including observation epoch MJD, EW, vmax, vmin, fdeep, the rest-frame time interval between S1 and S2, and log(Pχ2 ). To assess possible residual absorption remaining in the trough region of S2, we set upper limits on EW using a χ2 fitting approach. We assume that the absorption remaining in S2 has the same profile as that found in S1, scaled by a constant multiplicative factor, ǫ, lying between 0 and 1. We first find the value of ǫ providing the best fit to the trough region in S2 (spectral regions containing NALs as described in Section 2.3.4 are excised in this analysis). We then set a 90% confidence upper limit on ǫ by incrementing it until χ2 increases by 2.71 from the best-fit value (see, e.g., Section 15.6.5 of Press et al. 1992), and we use the scaled profile with this upper limit on ǫ to compute an upper limit on EW. Our EW limits are typically < 1.3 A˚ with a median upper limit of 0.8 A.˚ We note that our upper limits are conservative in that we do not allow for any narrowing of the absorption profile as it becomes weaker; analysis of our full BAL sample shows that such narrowing is typical. 17 Observed Wavelength 4600 4800 5000 5200 5400 5600 λ 2.0 J004022.40+005939.6 z =2.565 1.5 ∆ t =819.36 days 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 4 2 σ 0 N −2 −4 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Observed Wavelength 3700 3800 3900 4000 4100 4200 4300 4400 4500 4600 4700 λ J074650.59+182028.7 z =1.9163 1.5 ∆ t =900.46 days 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 5 σ 0 N −5 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Observed Wavelength 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500 4600 λ 2.0 J081102.91+500724.2 z =1.8422 ∆ t =1268.38 days 1.5 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 2 σ 0 N −2 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Fig. 2.3: Multi-epoch spectra of quasars with disappearing BAL troughs obtained from SDSS-I/II (red) and BOSS (black). The x-axes show observed (top) and rest-frame (bot- tom) wavelengths in A.˚ The dashed vertical lines show the emission-line rest wavelengths of Si iv λλ 1393, 1402 A˚ and C iv λλ 1548, 1550 A.˚ BALs are shown as shaded areas, and the red and black horizontal lines indicate BALs in the same color as the corresponding spectra. Solid blue bars show disappearing BAL troughs, and dashed blue bars show wavelengths of corresponding Si iv velocities. Quasars with a green star next to their names are the ones that are identified as “pristine” examples of BAL disappearances (see Section 2.3.3). The lower section of each panel shows the Nσ values for SDSS vs. BOSS observations (black). The dashed horizontal lines show the 1σ level. ± 18 Observed Wavelength 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500 λ 2.0 J085904.59+042647.8 z =1.8104 ∆ 1.5 t =935.10 days 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 2 σ 0 N −2 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Observed Wavelength 4400 4600 4800 5000 5200 5400 λ J093418.28+355508.3 z =2.4402 1.5 ∆ t =754.03 days 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 2 σ 0 N −2 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Observed Wavelength 3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 λ 2.0 J093620.52+004649.2 z =1.7213 1.5 ∆ t =1193.11 days 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 5 σ 0 N −5 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Observed Wavelength 3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 λ J094806.58+045811.7 z =1.7371 1.5 ∆ t =1076.69 days 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 4 2 σ 0 N −2 −4 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Figure 2.3–Continued 19 Observed Wavelength 3800 4000 4200 4400 4600 4800 λ J104841.02+000042.8 z =2.0263 1.5 ∆ t =1209.73 days 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 10 σ 0 N −10 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Observed Wavelength 3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500 λ J112602.81+003418.2 z =1.7928 1.5 ∆ t =1418.29 days 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 5 σ 0 N −5 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Observed Wavelength 3800 3900 4000 4100 4200 4300 4400 4500 4600 4700 4800 λ 2.0 J114546.22+032251.9 z =2.0075 ∆ t =1224.61 days 1.5 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 4 2 σ 0 N −2 −4 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Observed Wavelength 3900 4000 4100 4200 4300 4400 4500 4600 4700 4800 4900 λ J132216.24+052446.3 z =2.0498 1.5 ∆ t =1067.94 days 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 5 σ 0 N −5 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Figure 2.3–Continued 20 Observed Wavelength 3400 3500 3600 3700 3800 3900 4000 4100 4200 4300 λ J133152.19+051137.9 2.0 z =1.7118 ∆ t =1207.32 days 1.5 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 5 σ 0 N −5 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Observed Wavelength 3900 4000 4100 4200 4300 4400 4500 4600 4700 4800 4900 λ J133211.21+392825.9 2.0 z =2.052 ∆ t =731.32 days 1.5 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 4 2 σ 0 N −2 −4 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Observed Wavelength 4400 4600 4800 5000 5200 5400 5600 λ J134544.55+002810.7 z =2.468 1.5 ∆ t =1063.15 days 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 5 σ 0 N −5 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Observed Wavelength 3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500 λ 2.0 J142132.01+375230.3 z =1.7791 ∆ t =954.99 days 1.5 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 4 2 σ 0 N −2 −4 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Figure 2.3–Continued 21 Observed Wavelength 4000 4200 4400 4600 4800 5000 λ J142140.27−020239.0 z =2.0878 1.5 ∆ t =949.87 days 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 5 σ 0 N −5 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Observed Wavelength 4200 4400 4600 4800 5000 5200 λ 2.0 J152149.78+010236.4 z =2.2386 ∆ 1.5 t =1223.37 days 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 4 2 σ 0 N −2 −4 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Observed Wavelength 3800 3900 4000 4100 4200 4300 4400 4500 4600 4700 4800 λ 2.5 J152243.98+032719.8 z =2.0002 2.0 ∆ t =1236.25 days 1.5 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 4 2 σ 0 N −2 −4 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Observed Wavelength 4400 4600 4800 5000 5200 5400 λ 2.0 J155119.14+304019.8 z =2.4104 ∆ 1.5 t =760.61 days 1.0 0.5 SDSS SiIV CIV BOSS Normalized Flux Density F 5 σ 0 N −5 1250 1300 1350 1400 1450 1500 1550 1600 Rest Wavelength Figure 2.3–Continued 22 We found that there are a few additional arguable cases of BAL disappearance that do not satisfy our formal definition for disappearing BAL troughs. Examples of such cases include J083817.00+295526.5, J092444.66 000924.1, J115244.20+030624.4, and − J131010.07+052558.8. The first three of these arguable cases arise because of difficulties in deciding if a BAL trough should be split into two distinct components (one of which disappears) or not; in such cases we follow the formal BAL definition criteria associated with Equation 1. The fourth case arises from a C iv BAL lying close to the Si iv emission line; in this case separation of continuum vs. line emission is difficult. We have investigated the Si iv absorption at velocities corresponding to disappearing C iv troughs. A total of 12 of 19 quasars have spectral coverage at the corresponding velocities in both the SDSS and the BOSS spectra. Two quasars (J004022.40+005939.6 and J081102.91+500724.2) show Si iv absorption at the corresponding velocity, and this absorption disappeared in the BOSS spectra for both of these quasars (see Figure 2.3). We have investigated the Mg ii and Al iii regions of the 19 quasars with disappearing BAL troughs. All BOSS spectra have full coverage of the Mg ii region, as do SDSS spectra in all but two cases (J004022.40+005939.6 and J134544.55+002810.7). None of the quasars with disappearing BAL troughs shows the presence of BALs in the Mg ii or Al iii regions in their SDSS or BOSS spectra. Thus, certainly 17 and likely all 19 of these quasars are high-ionization BAL quasars. Figure 2.4 shows the spectra of the six quasars with more than two epochs of observa- tion: J004022.40+005939.6, J081102.91+500724.2, J085904.59+042647.8, J132216.24+052446.3, J134544.55+002810.7, and J155119.14+304019.8. Note the con- sistency of the BOSS spectra across the multiple epochs, which confirms the BAL disap- pearance and the general robustness of our analyses. 2.3.4 Notes on Specific Objects Here we describe the objects with remarkable properties or complex absorption spectra. Our descriptions supplement the objective approach adopted in Section 2.3.3 given the previously noted challenges sometimes associated with quantifying BAL troughs. J074650.59+182028.7: Both of the NAL doublets that are blended with the dis- appearing C iv BAL trough are identified as C iv absorption in the BOSS NAL database (Lundgren & York 2014 in preparation). J085904.59+042647.8: The NAL doublet that is blended with the disappearing C iv BAL trough is identified as C iv absorption (Lundgren & York 2014 in preparation). Figure 2.4 shows the three available epochs of observation for this quasar where the NAL appears to vary between two BOSS spectra (corresponding to a significance level of 95%) separated by 92 days in the rest frame. This apparent variability, combined with the location of this NAL within a BAL, suggest that the NAL is not an intervening absorption system at a lower redshift than the quasar but arises in gas outflowing from the quasar. 23 J004022.40+005939.6 J081102.91+500724.2 MJD 52261 MJD 51885 λ MJD 55182 ∆t = 819 days MJD 51912 ∆t = 9.5 days MJD 55186 Normalized Flux Density F MJD 55517 ∆ t = 821 days ∆t = 1278 days MJD 55444 ∆t = 893 days MJD 55590 ∆t = 1304 days 1360 1380 1400 1420 1440 1460 1480 1500 1520 1540 1560 1360 1380 1400 1420 1440 1460 1480 1500 1520 1540 1560 J085904.59+042647.8 J132216.24+052446.3 λ MJD 52649 MJD 52376 MJD 55277 MJD 55633 ∆t = 935 days ∆t = 1068 days Normalized Flux Density F MJD 55535 MJD 55703 ∆ ∆t = 1027 days t = 1091 days 1360 1380 1400 1420 1440 1460 1480 1500 1520 1540 1560 1340 1360 1380 1400 1420 1440 1460 1480 1500 1520 1540 1560 J134544.55+002810.7 J155119.14+304019.8 MJD 51666 λ MJD 53145 MJD 55739 MJD 51943 ∆t = 760 days ∆t = 80 days Normalized Flux Density F MJD 55748 ∆ MJD 55630 t = 763 days ∆t = 1143 days 1360 1380 1400 1420 1440 1460 1480 1500 1520 1540 1560 1340 1360 1380 1400 1420 1440 1460 1480 1500 1520 1540 1560 Rest Wavelength Rest Wavelength Fig. 2.4: Multi-epoch observations of the six quasars with disappearing BAL troughs that have more than one SDSS or BOSS spectrum. Continuum-normalized SDSS (red) and BOSS (black) spectra are shown ordered by observation date. MJD values for each ob- servation and the rest-frame time interval since the first observation are indicated under each spectrum. Horizontal dashed lines show continuum levels for each spectrum, the thick marks on the y-axis show the zero level for each spectrum, and shaded areas indicate BAL troughs. 24 J093418.28+355508.3: The SDSS spectrum of this quasar shows an adjacent ab- sorption feature blueward (between rest-frame 1416–1421 A)˚ of the disappearing C iv BAL trough. This feature could be a distinct mini-BAL trough or the continuation of a BAL trough that was truncated in our formal approach owing to statistical fluctuations. Since this feature also disappears in the BOSS spectrum, its nature does not affect the interpre- tation of a BAL disappearance. In the BOSS spectrum there is weak residual absorption that fails our formal defi- nition (see Section 2.3.2 and especially Equation 1) for being a BAL or mini-BAL. J093620.52+004649.2: According to Lundgren et al. (2007), the C iv BAL trough of this object emerged over a period of 105.6 days in the rest frame between two SDSS ≤ observations; the possible C iv absorption present in the first SDSS epoch does not qualify as a BAL. We now find that this same trough has disappeared over a period of 3.3 yr, and ≤ no remaining C iv absorption of any kind is apparent in the BOSS spectrum. Figure 2.5 shows the C iv BAL trough region in all three epochs. To our knowledge, this is the only reported example of a combined emergence and disappearance event for a BAL trough. 30 J093620.52+004649.2 MJD 52027 20 10 λ 30 MJD 52314 ∆ t = 105.46 days 20 10 Flux Density F 40 MJD 55563 ∆ t = 1299.38 days 30 20 1400 1420 1440 1460 1480 1500 1520 1540 1560 1580 Rest Wavelength Fig. 2.5: Three-epoch observations of J093620.52+004649.2 that show the only reported example of a combined emergence and disappearance event for a BAL trough. The BAL trough of this object emerged between two SDSS observations (top and middle panels; Lundgren et al. 2007) and disappeared in the BOSS spectrum (bottom panel). The dashed red lines show the BAL trough region. J094806.58+045811.7: The disappearing BAL trough in this case is clearly dis- tinct from the non-disappearing trough at higher velocity. The two troughs are separated 25 by 22 pixels ( 1500 km s−1), 12 of which lie above the 90% continuum threshold (see ≈ Section 2.3.2). In fact, 6 pixels even lie above the continuum level. J104841.02+000042.8: The NAL doublet that is blended with the weak disappear- ing C iv BAL trough is identified as C iv absorption (Lundgren & York 2014 in preparation). J112602.81+003418.2: In the BOSS spectrum there is weak residual absorption that fails our formal definition for being a BAL or mini-BAL. J132216.24+052446.3: In the BOSS spectrum there is weak residual absorption that fails our formal definition for being a BAL or mini-BAL. J133152.19+051137.9: The disappearing C iv BAL trough is formally separated from the adjacent BAL trough at lower outflow velocities according to our BAL identi- fication criteria. However, it is possible that there is some degree of physical connection between these two absorption systems. If they are indeed significantly connected, this would undermine the evidence for BAL disappearance in this quasar. It is notable, as apparent from the Nσ residuals in Figure 2.3, that the the lower velocity BAL trough does not vary significantly as the higher velocity BAL trough disappears; this is suggestive of a lack of physical connection between these two systems. Further observations of this quasar are required for clarification. J133211.21+392825.9: In the BOSS spectrum there is weak residual absorption that fails our formal definition for being a BAL or mini-BAL. J134544.55+002810.7: This is the most complex, as well as the strongest, absorp- tion system in our sample of disappearing BALs. In the SDSS spectrum, the disappearing BAL trough appears to be attached to a strong higher velocity C iv NAL doublet (see Lundgren & York 2014 in preparation) as well as a lower velocity mini-BAL. After the disappearance of the BAL trough, the NAL remains while the mini-BAL transforms to a NAL. Even if we consider the NAL, BAL and mini-BAL together as a connected BAL complex, the complex would still be listed as a disappearing BAL (since the only remain- ing features in the BOSS spectrum are NALs; see Section 3.3). We also note that there is another potentially connected mini-BAL at 1470–1485 A˚ in the SDSS spectrum; this mini-BAL disappears in the BOSS spectrum. J142132.01+375230.3: In the BOSS spectrum there is weak residual absorption that fails our formal definition for being a BAL or mini-BAL. J142140.27 020239.0: In the BOSS spectrum there is weak residual absorption − that fails our formal definition for being a BAL or mini-BAL. J152149.78+010236.4: In the BOSS spectrum there is weak residual absorption that fails our formal definition for being a BAL or mini-BAL. 26 2.4 Statistical Properties of Disappearing BALs 2.4.1 How Common is BAL Disappearance? The disappearance of BAL troughs is expected to be a rare event (e.g., Gibson et al. 2008; Hall et al. 2011). We have investigated 925 distinct BAL troughs in 582 quasars observed over 1.1–3.9 yr in the rest frame. From this sample, we have identified Ndt = 21 disappearing BAL troughs in Nqdt = 19 distinct quasars. Thus, on the observed +0.6 timescales, fdisapp = 2.3−0.5% (i.e., 21/925) of BAL troughs disappear, and fquasar = +0.9 3.3−0.7% (i.e., 19/582) of BAL quasars show a disappearing trough (quoted error bars are at 1σ confidence following Gehrels 1986). If we consider only the pristine-sample objects defined in Section 2.3.3, the observed fraction of disappearing C iv BAL troughs is +0.6 fdisapp = 1.2−0.5% (i.e., 11/925) and the fraction of BAL quasars showing a disappearing +0.8 trough is fquasar = 1.9−0.6% (i.e., 11/582). Table 2.3 lists the stated percentages above and the other quantities introduced below both for the standard sample and the pristine sample. One potential implication of the observed frequency of disappearing C iv BAL troughs is a relatively short average trough rest-frame lifetime t¯ ∆t /f = 109+31 yr, where ∆t is the average of the maximum time trough ≈ h maxi disapp −22 h maxi between observation epochs in a sample of BAL quasars (2.5 yr in this case) and fdisapp is the fraction of troughs that disappear over that time (also see Table 2.3). Here we state the average trough lifetime and average maximum observation time with different notations indicating that t¯ is inferred, whereas ∆t is measured. Note that by lifetime we trough h maxi mean the time over which a trough is seen along our line of sight and not necessarily the lifetime of the gas clouds responsible for the absorption. The BAL phenomenon can last longer than t¯trough if troughs come and go along our line of sight within BAL quasars. If many troughs have extremely long lifetimes then the above is only a lower limit on the true t¯trough. In that case, the fraction fshort of relatively short-lived troughs that dominate the parent population of disappearing troughs must have lifetimes of only fshort t¯trough. Thus, we conclude that a significant fraction of BAL troughs have average lifetimes of a century or less. Notably, our sample includes Ntransform = 10 examples of objects that have appar- ently transformed from BAL to non-BAL quasars; these are denoted in Table 2.1. BOSS spectra of these objects demonstrate that their C iv and Si iv regions do not have any re- maining BAL (or mini-BAL) troughs.3 Only two of these 10 objects (J004022.40+005939.6 and J152149.78+010236.4) have BOSS spectral coverage of the corresponding Lyα and N v BAL transitions as well, and visual inspection confirms that these spectra do not show any 3Two of these objects, J114546.22+032251.9 and J152149.78+010236.4, do have strong multiple NAL systems in their BOSS spectra that do not overlap in velocity with the disappearing C iv BAL trough. 27 remaining BAL or mini-BAL troughs in Lyα or N v (see Figure 2.6).4 As we are not aware of any BAL quasars that show convincing Lyα or N v absorption without also showing C iv or Si iv absorption, we consider the lack of remaining absorption in the latter transitions alone to be credible evidence of transformation from a BAL to a non-BAL quasar.5 The one qualification here is that it is possible, in principle, that some of these objects could have transitioned to a very high-ionization BAL state like that seen in SBS 1542+541 (Telfer et al. 1998); SBS 1542+541 shows strong O vi absorption but only weak or no absorption in transitions like C iv, Si iv, and N v. However, we consider this an unlikely possibil- ity because such very high-ionization BAL quasars appear to be rare based on a visual search of the spectra of high-redshift SDSS DR7 catalog (Schneider et al. 2010) quasars (by P.B.H.) and on their absence as contaminants in searches for damped Lyα absorption systems in the same catalog (J.X. Prochaska 2011, personal communication). The fraction +0.7 of BAL quasars transforming to non-BAL quasars is ftransform = 1.7−0.5% (i.e., 10/582) on 1.1–3.9 yr timescales. These are the first reported examples of BAL to non-BAL quasar transformations; the four previously reported BAL disappearance events described in Sec- tion 2.1 did not involve transformations from BAL to non-BAL quasar status (i.e., troughs from other ions or additional C iv troughs remained in the spectra). These BAL to non-BAL quasar transformations set a lower limit on the lifetime of the ultraviolet BAL phenomenon along our line of sight of t¯ ∆t /f = 150+60 yr. However, defining the lifetime of the BAL phase BAL ≃ h maxi transform −50 along our our line of sight is more complicated than defining a trough lifetime along our line of sight. For example, if a BAL quasar has its only trough disappear but then has a different trough appear a year later, should this event be considered separate BAL phases or one phase with patchy absorption? Monitoring the strong X-ray absorption which is also characteristic of BAL quasars (e.g., Gallagher et al. 2002, 2006) would improve our understanding of how connected UV absorption variability is to absorption variability at other wavelengths, and thus to variability between BAL and non-BAL phases along a given line of sight, all of which can help test BAL outflow models. 2.4.2 Luminosities, Black-Hole Masses, Reddening, and Radio Properties of Quasars with Disappearing BAL Troughs In Figure 2.7, we compare our main-sample quasars to quasars with disappearing troughs in a plot of redshift versus absolute i-band magnitude (Mi). We physically expect 4The BOSS spectrum of J133211.21+392825.9 shows intervening absorption from a damped Lyα absorber that is not related to the BAL phenomenon. The intervening nature of this absorption is supported by its narrow width and symmetric shape, as well as by the fact that the maximum depth of the line is consistent with zero flux. 5One possible counterexample is the remarkable quasar PDS 456 which may show only broad Lyα absorption (O’Brien et al. 2005). However, the identification of the single broad absorption feature found in the spectrum of this quasar is not clear; e.g., it could be from highly blueshifted C iv. 28 Observed Wavelength 4000 4500 5000 5500 6000 6500 7000 3.0 SDSS λ J004022.40+005939.6 BOSS 2.5 2.0 1.5 1.0 Normalized Flux Density F 0.5 1100 1200 1300 1400 1500 1600 1700 1800 1900 Rest Wavelength Observed Wavelength 3500 4000 4500 5000 5500 6000 3.5 SDSS λ 3.0 J133211.21+392825.9 BOSS 2.5 2.0 1.5 1.0 Normalized Flux Density F 0.5 1100 1200 1300 1400 1500 1600 1700 1800 1900 Rest Wavelength Observed Wavelength 3500 4000 4500 5000 5500 6000 SDSS 3.0 λ J152149.78+010236.4 BOSS 2.5 2.0 1.5 1.0 Normalized Flux Density F 0.5 1100 1200 1300 1400 1500 1600 1700 1800 1900 Rest Wavelength Fig. 2.6: BAL quasars J004022.40+005939.6, J133211.21+392825.9, and J152149.78+010236.4 that transformed to non-BAL quasars after the disappearance of the C iv BAL trough (marked with a blue bar). Green bars under each spectrum indicate the velocities in the Lyα, N v, and Si iv regions corresponding to the disappearing C iv trough. Vertical dashed lines indicate the emission lines for Lyα, N v, Si iv,C iv, and C iii, respectively, from left to right. 29 that, since the sampled rest-frame timescales are shortened toward higher redshifts, less BAL variability will be seen with increasing redshift. Therefore, we applied the two-sample Kolmogorov-Smirnov (K-S) test to compare Mi distributions for quasars with disappear- ing BAL troughs and the main-sample quasars sampled on similar rest-frame timescales (i.e., z 2.6). We find no evidence that BAL quasars with disappearing troughs have ≤ exceptional Mi values compared to the main-sample quasars with similar redshifts (K-S probability of 68%). In addition, we compared SMBH mass estimates from Shen et al. (2011) for quasars with disappearing BAL troughs and our main-sample quasars. The quasars with disappearing C iv BAL troughs do not show any remarkable inconsistency from the main sample (K-S probability of 9% for z 2.6 quasars). Thus, the disap- ≤ pearing trough phenomenon appears to be generally present throughout the BAL quasar population. −29 ) i −28 Absolute i−Band Magnitude (M −27 Main Sample Quasars with a disappearing BAL trough −26 2 2.5 3 3.5 4 4.5 Redshift (z) Fig. 2.7: Redshifts vs. absolute i-band magnitudes of main-sample quasars in this study (black solid circles) and quasars with disappearing C iv BAL troughs (red triangles). Red- shifts are from Hewett & Wild (2010), and Mi values are from Shen et al. (2011). We have also investigated if our quasars with disappearing C iv BAL troughs show any difference in intrinsic reddening from the main sample. Following G09, we have calcu- lated a basic “reddening parameter”, defined as the ratio of the continuum flux densities at 30 1400 A˚ and 2500 A.˚ We do not detect any difference in the distributions of this parameter for the quasars with disappearing C iv BAL troughs and the main sample. The FIRST survey has detected radio emission from two of the quasars with dis- appearing troughs. SDSS J004022.40+005939.6 is radio intermediate with R = 60.7 and J081102.91+500724.2 is radio loud with R = 223.8. The quasars without detected radio emission have R< 5 and are thus radio quiet. The BAL variability of radio-loud BAL quasars is not well understood and is just now being studied systematically (e.g., Miller et al. 2012). Our results for J004022.40+005939.6 and J081102.91+500724.2 demonstrate that BAL disappearance is a phenomenon of both radio-loud and radio-quiet quasars. 2.4.3 EWs, Depths, Velocities, and Widths of Disappearing BAL Troughs We have investigated the characteristics of disappearing C iv BAL troughs by com- paring them with non-disappearing ones. Figure 2.8 shows a comparison of the EW dis- tributions for disappearing BAL troughs, the other BAL troughs present in quasars that show one disappearing trough (i.e., those that do not disappear; see Section 2.4.4 for fur- ther discussion), and all 925 distinct BAL troughs in the main sample. In order to be consistent, we measured the BAL-trough parameters from the observations obtained at epoch t1 (i.e., the latest observation from the SDSS). The K-S test results show that there is only a 0.09% chance of consistency between the EW distributions of disappearing BAL troughs and all main-sample BAL troughs. Based on this result and Figure 2.8, it ap- pears that BAL disappearance tends to occur for weak or moderate-strength absorption troughs but not the strongest ones. In particular, no troughs with EW > 12 A˚ disappeared. Furthermore, the K-S test results comparing the main-sample C iv BAL troughs and the additional non-disappearing C iv BAL troughs defined in Section 2.4.4 show no evidence for inconsistency (K-S probability of 56.9%). 25 Figure 2.9 shows the distributions of the BAL-trough depth parameter, fdeep, for disappearing BAL troughs and for all 925 distinct BAL troughs in the main sample. The K-S test shows that these two distributions only have a 0.03% chance of consistency. The BAL troughs that disappear are shallower than BAL troughs in general, although some fairly deep BAL troughs do disappear. Figures 2.10a and 2.10b show the distributions of vmax and vmin (see Section 2.3.2) iv for disappearing and all 925 distinct C BAL troughs in the main sample. The vmax distribution of all BALs shows that this quantity is roughly distributed evenly between −1 3000 and 25000 km s , while the corresponding vmin distribution rises toward low − − 1 velocities. The peak in the v distribution around 3000 km s− is artificially elevated due min − to the assigned lower velocity limit. The distribution histograms in Figures 2.10a and 2.10b show basic agreement with the similar distributions presented in G09 (see their Figure 8), although the vmax and vmin values in G09 give the velocities for all BALs together in a given spectrum. A K-S test for the vmax distributions of disappearing BAL troughs and all main- sample BAL troughs shows that these two distributions are not demonstrably inconsistent 31 8 6 Disappearing BAL Troughs 4 2 4 Other BALs for Quasars with DisappearingTrough 2 Number of BAL Troughs 150 100 BAL Troughs 50 0 0 10 20 30 40 50 60 EW ˚ Tr ( A) Fig. 2.8: EW distributions for disappearing C iv BAL troughs (upper panel, red), the other BAL troughs present in quasars that show one disappearing trough (middle panel, green), and all 925 distinct C iv BAL troughs in the main sample (lower panel, blue). (K-S probability of 11.1%), while the corresponding test using the vmin distributions gives a probability of 0.1%. Based on this result and Figures 2.10a and 2.10b, it appears that BAL disappearance tends to occur for BAL troughs with relatively high values of vmin; this result indicates that our requirement of trough detachment (see Section 2.2.2) should not cause significant biases to our sample statistics. Figures 2.10c and 2.10d show the distributions of the central velocity [v = (v + v )/2] and trough width (∆v = v v ) for disappearing and c max min | max − min | all 925 distinct C iv BAL troughs in the main sample. A K-S test yields a probability of 1.5% for consistency between the two vc distributions, indicating that disappearing BALs tend to have higher outflow central velocities than BALs in general. A K-S test for the ∆v distributions of disappearing BAL troughs and all main-sample BAL troughs shows that these two distributions have a 6.9% chance of consistency. These distributions suggest that disappearing BAL troughs tend to be narrower than the average BAL troughs. According to our statistical tests above, C iv BAL disappearance generally occurs for weaker troughs as well as higher velocity troughs. It is possible that these two results are related, or that one is simply the effect of the other. For example, in our main sample we find that weak BAL troughs can generally achieve higher velocities than strong BAL 32 5 4 Disappearing BAL Troughs 3 2 1 140 120 Number of BAL Troughs 100 BAL Troughs 80 60 40 20 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 f25 deep 25 iv Fig. 2.9: Distribution of fdeep for disappearing C BAL troughs (upper panel) and for all 925 distinct C iv BAL troughs in the main sample (lower panel). troughs (although there is a wide range of velocity observed at all trough strengths); as a result, the average velocity for the population of weak BAL troughs is higher than that for strong BAL troughs. This basic result has also been noted by others (e.g., G09; Capellupo et al. 2011). Thus, the tendency for BAL disappearance to occur for weaker troughs could also lead to disappearing troughs having higher velocities, on average, than for the whole trough population. A larger sample of disappearance events will be required to determine if BAL strength or BAL velocity, if either, is primarily connected to BAL disappearance. 2.4.4 Connections Between Disappearing and Non-Disappearing Troughs Multiple troughs in the quasar spectra that have at least one disappearing BAL can be used to investigate connections between disappearing and non-disappearing troughs. Among the 19 quasars with disappearing troughs, we find nine that have at least one addi- tional C iv BAL trough that does not disappear ( J093418.28+355508.3, J094806.58+045811.7, J104841.02+000042.8, J112602.81+003418.2, J132216.24+052446.3, J133152.19+051137.9, J134544.55+002810.7, J142140.27 020239.0, and − J155119.14+304019.8). These nine quasars contain a total of 12 C iv troughs that do not disappear and satisfy the criteria in Section 2.2.2. We will hereafter refer to these as the “additional non-disappearing C iv troughs”. Figure 2.11 compares the EWs at two 33 5 4 Disappearing BAL Troughs a Disappearing BAL b 4 Troughs 3 3 2 2 1 1 100 250 80 200 BAL Troughs BAL Troughs Number of BAL Troughs 60 Number of BAL Troughs 150 40 100 20 50 0 0 −30000 −25000 −20000 −15000 −10000 −5000 −30000 −25000 −20000 −15000 −10000 −5000 −1 v (km s−1) v (km s ) max min 12 5 c d Disappearing BAL Troughs 10 4 8 Disappearing BAL Troughs 3 6 2 4 1 2 0 300 100 BAL Troughs 250 BAL Troughs Number of BAL Troughs Number of BAL Troughs 80 200 60 150 40 100 20 50 0 0 −30000 −25000 −20000 −15000 −10000 −5000 0 5000 10000 15000 20000 v (km s−1) −1 c ∆v (km s ) iv Fig. 2.10: vmax (a), vmin (b), vc (c), and ∆v (d) distributions for disappearing C BAL troughs (upper panels), and all 925 distinct C iv BAL troughs in the main sample (lower panels). 34 iv epochs, t1 and t2, for distinct C BAL troughs in the main sample and the additional non- disappearing C iv troughs. The additional non-disappearing C iv troughs almost always weaken (by up to 85%), except for the least blueshifted trough of J133152.19+051137.9, which strengthens by 7%. In J094806.58+045811.7, J104841.02+000042.8, J134544.55+002810.7, and J142140.27 020239.0, the additional non-disappearing C iv trough transforms from a − BAL to a mini-BAL as it weakens. The weakening of the additional non-disappearing C iv troughs is statistically significant, given that the combinatorial probability to have 11 or more troughs out of 12 weaken is only 0.3%. The fact that the additional non-disappearing troughs usually weaken indicates that variability across multiple troughs is coordinated. 25 4× 2× 20 ˚ A) 15 ( ) 2 EW(t 10 0.5× 5 0.25× 0 0 5 10 15 20 25 EW (t ) 1 ( A)˚ Fig. 2.11: The EWs at two epochs for all BAL troughs in the main sample (grey circles). The green triangles mark the other BAL troughs present in quasars that show one disap- pearing trough; note that in all cases but one the other BAL troughs weaken. Dashed lines indicate four times, two times, one-half of, and one-quarter of the first-epoch EWs, from top to bottom. Figure 2.12 shows fractional EW variations for the 12 additional non-disappearing C iv troughs versus central velocity offset from the disappearing trough. The error bars on EW fractional changes are calculated from the EW uncertainties at the two epochs (i.e., t1 and t2). Notably, the coordinated variability across multiple troughs persists even for velocity offsets as large as 10000–15000 km s−1. This result is even more remarkable given that BAL troughs often vary in discrete regions only a few thousand km s−1 wide (Gibson et al. 2008), and it requires further investigation. 35 −1.6 −1.4 −1.2 75% −1.0 > Tr −0.8 / Tr −0.6 50% EW ∆ −0.4 Faster Troughs Slower Troughs −0.2 0 0% 15000 10000 5000 0 −5000 −10000 −15000 Velocity Offset from Disappearing Trough (km s −1) Fig. 2.12: Fractional changes in EW for 12 additional non-disappearing C iv BAL troughs present in quasars that show one disappearing trough. The x-axis is the central veloc- ity [(vmax + vmin)/2] of each trough relative to that of the trough that disappeared. The horizontal dashed red lines indicate the levels for 50%, 75%, and 0% decrease in trough strength. Note that even BAL troughs separated by 10000–15000kms−1 from the disap- pearing troughs weaken, showing coordinated variations over a large velocity range. 36 In Figure 2.12 nine of the 12 additional non-disappearing C iv troughs have smaller velocities than the disappearing C iv troughs. This result implies that, for BAL quasars showing multiple troughs, the highest velocity trough is usually the one that disappears. This suggestive result is not clearly significant in the current sample, with a combinatorial probability of chance occurrence of 7%, and it needs further investigation. We also investigated Si iv BAL troughs at velocities that correspond to one of the additional non-disappearing C iv troughs. The additional Si iv BAL troughs that are found in the spectra of three quasars show dramatic variations; the Si iv BAL troughs in J132216.24+052446.3 and J142140.27 020239.0 transform to mini-BALs (weakening in − EW by 88% and 87%, respectively), and the trough in J155119.14+304019.8 disappears. The results above demonstrate that outflow stream lines separated by large veloc- ities apparently can either coordinate their variability or are being acted upon by some external agent that enforces coordination. The latter possibility appears more likely given understanding of BAL winds. One example of an external agent could be disk rotation. Such rotation might lead to coordinated observed changes in absorption by several non- axisymmetric outflows that are loosely anchored to the accretion disk at different radii, provided there is some large-scale azimuthal asymmetry of the disk similarly affecting several stream lines. Another possible external agent could be the “shielding gas” that prevents BAL outflows from being overionized by highly energetic emission generated close to the SMBH (e.g., Murray et al. 1995; Proga et al. 2002). This shielding gas is found to be variable in wind simulations (e.g., Sim et al. 2010), and the limited long-term X-ray variability observations show rare apparent examples of shielding-gas variations (e.g., Saez et al. 2012). Variations of the shielding gas could change the level of ionizing-continuum radiation reaching larger radii. In response to these changes in ionizing continuum, several absorption components at different velocities could rise and fall in ionization level together with some features disappearing. One challenge for this scenario is that BAL-trough varia- tions generally do not appear to be correlated with variations of the observable continuum (see Section 2.1). However, since the observable continuum is typically that longward of Lyα, it is perhaps possible that the ionizing continuum at shorter wavelengths varies differently in at least some cases. 2.5 Summary and Future Work We have used a systematically observed sample of 582 BAL quasars with 925 dis- tinct C iv BAL troughs to provide the first statistically meaningful constraints upon BAL disappearance on multi-year timescales. Our main results are the following: 1. We have identified 21 cases of C iv BAL disappearance in 19 quasars. On rest- +0.6 frame timescales of 1.1–3.9 yr, fdisappear = 2.3−0.5% of BAL troughs disappear and +0.9 fquasar = 3.3−0.7% of BAL quasars show a disappearing trough. If we consider only 37 the pristine sample defined in Section 2.3.3, then we find 11 cases of C iv BAL disap- +0.4 pearance in 11 quasars; the corresponding percentages are fdisappear = 1.2−0.4% and +0.8 fquasar = 1.9−0.6%. See Section 2.4.1. 2. The observed frequency of disappearing C iv BAL troughs suggests an average trough rest-frame lifetime of 100–200 yr. See Section 2.4.1. 3. Ten quasars showing C iv BAL disappearance have apparently transformed from BAL to non-BAL quasars; these are the first reported examples of such transformations. The frequency of BAL to non-BAL quasar transformation on timescales of 1.1–3.9 yr +0.7 is 1.7−0.5%. See Section 2.4.1. 4. The BAL quasars with disappearing troughs have representative luminosities (Mi values), SMBH mass estimates, and intrinsic reddening compared to our sample as a whole. See Section 2.4.2. 5. As expected from the fact that most BAL quasars are radio quiet, most BAL quasars with disappearing troughs are radio quiet. However, we do find one such quasar that is radio loud and another that is radio intermediate. Thus, BAL disappearance is a phenomenon of both radio-quiet and radio-loud quasars. See Section 2.4.2. 6. BAL disappearance appears to occur mainly for weak or moderate-strength absorp- tion troughs but not the strongest ones; e.g., no troughs with EW > 12 A˚ disap- peared. The BAL troughs that disappear are shallower than BAL troughs in general, although some fairly deep BAL troughs do disappear. See Section 2.4.3. 7. Disappearing C iv BAL troughs show higher outflow velocities than BAL troughs in general, as indicated by their measured central velocities and minimum velocities (though their measured maximum velocities do not appear exceptional). There is also suggestive evidence that disappearing BAL troughs tend to be narrower than BAL troughs in general. This tendency for BAL disappearance to occur for higher velocity troughs could be related to, or even a secondary effect of, the fact that BAL disappearance appears to occur mainly for weak or moderate-strength absorption troughs (see point 5 above). See Section 2.4.3. 8. When one BAL trough in a quasar spectrum disappears, the other present troughs usually weaken (11 times out of 12 in our sample, corresponding to a significance level of > 99%). The phenomenon occurs even for velocity offsets as large as 10000–15000 km s−1. Variability across multiple troughs appears surprisingly coor- dinated. Possible causes of such coordinated variations could be disk-wind rotation or variations of shielding gas that lead to variations of ionizing-continuum radiation. These possible agents will need to be considered in future models of quasar winds. See Section 2.4.4. 38 Given the results above, we can identify several promising observational projects that should extend understanding of BAL disappearance. Further spectroscopy of the quasars that have shown BAL disappearance will allow a search for reappearance of any of these BALs. Such reappearances at the same measured velocities would not be ex- pected if wind stream lines have moved out of the line of sight owing to rotation of a non-axisymmetric outflow. However, BAL reappearance should be possible if the disap- pearance is a consequence of BAL weakening to strengths below our detection threshold. Systematic large-sample variability studies should let us assess the extent to which BAL disappearance is just the extension of normal BAL variability down to very small EWs. Further spectroscopy will also allow monitoring of the additional non-disappearing troughs. Furthermore, the planned absolute flux calibration of the BOSS spectra (e.g., Margala & Kirkby 2011) will allow a search for any systematic continuum-level changes associated with BAL disappearance. The rate of BAL emergence events must balance that of BAL disappearance events if the BAL quasar population is in a steady state, and thus systematic large-scale studies of BAL emergence will be a critical complement to those of disappear- ance. Finally, multiwavelength observations of the quasars showing BAL disappearance are worthwhile. For example, X-ray observations of objects that have transformed from BAL to non-BAL quasars will be able to assess if the X-ray absorbing shielding gas is still present along the line of sight. The main-sample data set utilized in this study, along with the still incoming BOSS observations, will be effective for a variety of additional investigations of BAL variability. These include studies of (1) absorption EW variability as a function of timescale for different BAL transitions, (2) connections between BAL, emission-line, and reddening variability, and (3) the effects of luminosity, redshift, SMBH mass, Eddington fraction, and radio properties on BAL variability. 39 Table 2.1. Sample of Quasars Showing Disappearing BAL Troughs a b c d ′e f SDSS Name Redshift i Mi Plate-MJD-Fiber BI NTr z (mag) (mag) (km s−1) J004022.40+005939.6‡ 2.565 0.0006 19.223 0.027 27.094 0690-52261-563 1668 1 ± ± − 3587-55182-950 0 0 3589-55186-558 0 0 4222-55444-710 0 0 J074650.59+182028.7†,‡ 1.9163 0.0005 18.043 0.016 27.584 1582-52939-095 1241 1 ± ± − 4492-55565-828 0 0 J081102.91+500724.2‡ 1.8422 0.0006 18.838 0.016 26.720 0440-51885-377 591 1 ± ± − 0440-51912-395 665 1 3699-55517-062 0 0 4527-55590-028 0 0 J085904.59+042647.8†,‡ 1.8104 0.0005 18.812 0.022 26.646 1192-52649-291 854 1 ± ± − 3817-55277-538 0 0 3814-55535-928 0 0 J093418.28+355508.3 2.4402 0.0007 18.902 0.016 27.400 1275-52996-096 1621 2 ± ± − 4575-55590-498 1007 1 J093620.52+004649.2‡ 1.7213 0.0005 18.391 0.016 27.001 0476-52314-444 958 1 ± ± − 3826-55563-542 0 0 J094806.58+045811.7 1.7371 0.0006 18.640 0.024 26.704 0994-52725-288 2146 3 ± ± − 4798-55672-934 317 1 J104841.02+000042.8† 2.0263 0.0006 18.720 0.016 26.970 0276-51909-310 2133 2 ± ± − 3835-55570-398 0 0 J112602.81+003418.2 1.7928 0.0005 18.082 0.016 27.348 0281-51614-432 2421 2 ± ± − 3839-55575-844 1024 1 J114546.22+032251.9‡ 2.0075 0.0007 19.058 0.023 26.721 0514-51994-458 389 1 ± ± − 4766-55677-050 0 0 J132216.24+052446.3 2.0498 0.0006 18.384 0.019 27.438 0851-52376-622 2903 4 ± ± − 4761-55633-794 1154 1 4839-55703-442 1125 1 J133152.19+051137.9 1.7118 0.0005 18.182 0.019 27.150 0852-52375-626 3840 3 ± ± − 4759-55649-756 3255 2 J133211.21+392825.9‡ 2.0520 0.0009 19.021 0.023 26.760 2005-53472-330 690 1 ± ± − 4708-55704-412 0 0 Continued on Next Page... 40 Table 2.1 Continued. a b c d ′e f SDSS Name Redshift i Mi Plate-MJD-Fiber BI NTr z (mag) (mag) (km s−1) J134544.55+002810.7 2.4680 0.0005 18.535 0.019 27.810 0300-51666-426 1712 1 ± ± − 0300-51943-382 2875 1 4043-55630-868 0 0 J142132.01+375230.3‡ 1.7791 0.0006 18.658 0.019 26.725 1380-53084-013 854 1 ± ± − 4712-55738-030 0 0 J142140.27 020239.0 2.0878 0.0006 18.877 0.016 27.044 0917-52400-546 3950 3 − ± ± − 4032-55333-736 1002 1 J152149.78+010236.4‡ 2.2386 0.0004 18.558 0.018 27.602 0313-51673-339 807 1 ± ± − 4011-55635-166 0 0 J152243.98+032719.8‡ 2.0002 0.0005 18.653 0.018 27.172 0592-52025-254 374 1 ± ± − 4803-55734-442 0 0 J155119.14+304019.8 2.4104 0.0004 18.493 0.016 27.826 1580-53145-008 5176 2 ± ± − 5011-55739-054 417 1 5010-55748-492 382 1 aRedshifts are from Hewett & Wild (2010), calculated from the cross correlation of the Mg ii, C iii], and C iv emission lines. bThe i-band magnitude given in the SDSS DR5 quasar catalog (Schneider et al. 2007). cAbsolute i-band magnitude from Shen et al. (2011). dUnique Plate-MJD-Fiber numbers for each spectrum. BOSS spectra have MJD 55176 (see Section 4 of Ross et al. 2012). ≥ eBalnicity index of each quasar in the given observation, summed over all troughs in the velocity range 3000 v 30000kms−1. None of the quasars in the main sample has a disappearing BAL trough −beyond≥ this velocity≥ − range. f Number of BAL troughs in each spectrum. †Blended NAL(s) with disappearing BAL troughs. ‡Quasars that transformed from BAL to non-BAL quasars. 41 Table 2.2. Parameters of Disappearing BAL Troughs a 25 b c d 2 Name MJD EW vmax vmin fdeep ∆t log(Pχ ) SDSS (A)˚ (kms−1) (kms−1) (days) J004022.40+005939.6 52261 10.60 0.705 10067 4167 0.74 819.35 < 300 † ± − − − J074650.59+182028.7 52939 8.76 0.212 24994 18015 0.46 900.46 < 300 J081102.91+500724.2 51912 4.70±0.477 −12405 − 9830 0.74 1268.38 −8.59 † ± − − − J085904.59+042647.8 52649 5.22 0.476 19315 16370 0.73 935.10 11.34 J093418.28+355508.3 52996 3.60±0.272 −25575 −21959 0.31 754.03 −13.79 J093620.52+004649.2 52314 6.15±0.373 −17603 −13677 0.68 1193.91 <− 300 J094806.58+045811.7 52725 5.98±0.549 −21138 −16797 0.55 1076.69 15− .96 † ± − − − J104841.02+000042.8 51909 2.25 0.185 12757 10654 0.33 1209.73 11.51 J112602.81+003418.2 51614 4.27±0.148 −26449 −22919 0.54 1418.29 <− 300 J114546.22+032251.9 51994 3.34±0.328 −12771 − 9682 0.28 1224.61 11− .60 J132216.24+052446.3 52376 2.72±0.184 −22727 −20322 0.38 1067.94 <− 300 52376 3.02±0.218 −18691 −15564 0.26 1067.94 −9.29 52376 2.30±0.176 −13360 −11213 0.33 1067.94 −9.90 J133152.19+051137.9 52375 3.16±0.219 −12549 −10440 0.56 1207.32 <− 300 J133211.21+392825.9 53472 4.68±0.349 −21377 −17861 0.53 731.32 12− .41 J134544.55+002810.7 51943 7.60±0.182 −11269 − 8096 0.92 1063.15 <− 300 J142132.01+375230.3 53084 5.15±0.343 −17005 −14193 0.76 954.99 < −300 J142140.27 020239.0 52400 4.55±0.177 −15968 −12922 0.57 949.87 < −300 J152149.78+010236.4− 51673 6.03±0.460 −23658 −18462 0.41 1223.37 < −300 J152243.98+032719.8 52025 2.77±0.203 −14622 −12300 0.34 1236.25 < −300 J155119.14+304019.8 53145 8.20±0.428 −23588 −17834 0.64 760.61 < −300 ± − − − aMJD of the observation used for BAL parameter measurements, which is taken to be the last SDSS observation that possesses the disappearing trough. bFraction of BAL bins which lie at least 25% under the continuum. cThe rest-frame time interval between the last observation that possesses the disappearing trough, given in column 2, and the first observation that shows disappearance. d Logarithm of χ2 probability which gives the probability of consistency between the SDSS and BOSS observations in the region limited by vmax and vmin (see Section 2.3.3). †Blended NAL(s) with disappearing BAL troughs. 42 Table 2.3. Observed Fractions and Lifetimes for Disappearing BAL Troughs a b c d e f g h Ndt fdisapp Nqdt fquasar Ntransform ftransform t¯trough t¯BAL (%) (%) (%) (yr) (yr) Standard +0.6 +0.9 +0.7 +31 +60 Sample 21 2.3−0.5 19 3.3−0.7 10 1.7−0.5 109−22 150−50 Pristine +0.5 +0.8 +0.7 +105 +105 Sample 11 1.2−0.4 11 1.9−0.6 7 1.2−0.4 208−60 208−80 aNumber of disappearing C iv BAL troughs bFraction of disappearing C iv BAL troughs cNumber of quasars showing a disappearing C iv BAL trough dFraction of quasars showing a disappearing C iv BAL trough eNumber of quasars that transformed from BAL to non-BAL quasars f Fraction of quasars that transformed from BAL to non-BAL quasars gAverage trough lifetime hLifetime of the BAL phenomenon along our line of sight 43 Chapter 3 Broad Absorption Line Variation on Multi-Year Timescales in a Large Quasar Sample 3.1 Introduction The high-velocity winds from quasars are important for several related reasons. First, these winds can significantly affect observed quasar properties via, e.g., ultravio- let (UV) line absorption, high-ionization line emission, optical/UV reddening, and X-ray absorption (e.g., Weymann et al. 1981; Turnshek 1988; Leighly 2004; Collin et al. 2006; Gallagher et al. 2002, 2006; Gibson et al. 2009; Richards et al. 2011). Second, wind absorp- tion lines are observed frequently, indicating that winds have a high covering factor and are a substantial part of quasar nuclear regions (e.g., Ganguly & Brotherton 2008; Gib- son et al. 2009; Allen et al. 2011). Third, winds might improve the efficiency of accretion onto the central supermassive black hole (SMBH) by removing angular momentum from the accretion disk (e.g., Emmering et al. 1992; Konigl & Kartje 1994). Finally, winds can evacuate gas from the host galaxy, perhaps shaping SMBH growth and galaxy evolution (e.g., Di Matteo et al. 2005; Chartas et al. 2009; Rupke & Veilleux 2011; Sturm et al. 2011; Borguet et al. 2013). The strongest absorption lines created by quasar winds are Broad Absorption Line (BAL) troughs with velocity widths greater than 2000 km s−1 and typical outflow velocities of 1000–30000 km s−1 (e.g., Weymann et al. 1991). Many BALs are believed to be formed in an equatorial wind that is launched from the accretion disk at 10–100 light days from the SMBH (e.g., Murray et al. 1995; Proga et al. 2000). If the BALs are formed in the vicinity of the launching region, then the timescale for wind material to cross the region of interest is about 1–10 yr, and this is a reasonable characteristic timescale over which flow structures might be expected to change. This is also the characteristic timescale for sig- nificant angular rotation of the accretion disk at the wind-launching radius. Assessments of the transverse velocities of BAL material indicate these are often comparable to the aforementioned outflow velocities (e.g., Capellupo et al. 2011; Hall et al. 2011), and char- acteristic variability timescales of years are again deduced for material moving transversely through our line-of-sight. Thus, studies of multi-year BAL variability can provide useful insights into the nature of quasar winds. The existence of BAL variability has been known for over two decades (e.g., Smith & Penston 1988; Turnshek et al. 1988; Barlow et al. 1992). Early investigations of this 44 phenomenon were generally single-object studies with 2–4 observational epochs, although Barlow (1993) performed an early spectroscopic monitoring survey of 23 BAL quasars. In recent years, systematic sample-based studies of BAL variability, investigating 5–30 objects, have become increasingly common (e.g., Lundgren et al. 2007; Gibson et al. 2008, 2010; Capellupo et al. 2011, 2012; Filiz Ak et al. 2012; Haggard et al. 2012; Miller et al. 2012; Vivek et al. 2012); see Table 3.1 for a summary of the basic properties of these samples. Sample-based studies have the advantage of allowing broadly applicable and statistically reliable conclusions about BAL variability to be drawn. BAL variability has been found to be a complex and diverse phenomenon. Generally, changes in the residual flux in portions of BAL troughs are observed, while detections of BAL acceleration/deceleration are much rarer (e.g., Vilkoviskij & Irwin 2001; Gibson et al. 2008, 2010; Capellupo et al. 2012). We have been using observations taken as part of the ongoing Baryon Oscillation Spectroscopic Survey (BOSS; Dawson et al. 2013) of the Sloan Digital Sky Survey-III (SDSS-III; Eisenstein et al. 2011) to perform the largest survey of multi-year BAL vari- ability to date (see Section 3.2.1 and Filiz Ak et al. 2012 for further description). Our final sample will include 2100 BAL quasars with high-quality spectra providing multi-year ≈ variability coverage in the rest frame; this size is about two orders of magnitude larger than other samples being used to investigate multi-year BAL variability (see Table 3.1). In Filiz Ak et al. (2012) we presented some first results from the survey focused on C iv BAL disappearance events. In this paper, we provide more general findings regarding the variability of C iv and Si iv BAL troughs, primarily on multi-year timescales but also extending to much shorter timescales. These findings are based upon 291 BAL quasars se- lected from our full sample to have particularly high-quality spectroscopic coverage of these troughs; 428 distinct C iv and 235 distinct Si iv troughs are utilized. Our overall approach is first to characterize systematically how BAL troughs vary on multi-year timescales and then to use this characterization to derive physical implications for BAL outflows. For example, our results provide insights into the radial distance where most BAL troughs are formed, the lifetimes of BAL troughs along our line of sight, the connection between BAL disappearance/emergence events and general BAL variability, and the driving mechanisms of BAL variability. In Section 3.2 we describe the observations and sample selection underlying this work, and in Section 3.3 we describe data preparation and analysis approaches. Our observational results on BAL variability are presented in Section 3.4. In Section 3.5 we provide a discussion of implications for quasar winds, and in Section 3.6 we present a summary and describe promising future avenues of relevant research. −1 −1 Throughout this work we use a cosmology with H0 =70 km s Mpc ,ΩM = 0.3, and ΩΛ = 0.7. All time intervals and EWs are in the rest frame of the quasar unless stated otherwise. Negative signs for velocities indicate that a BAL trough is blueshifted with respect to the systemic velocity. We define EWs of absorption features to be positive. Positive EW variations indicate strengthening, and negative values indicate weakening. 45 3.2 Observations and Sample Selection 3.2.1 Observations We have utilized data from the Sloan Digital Sky Survey-I/II (hereafter “SDSS”; York et al. 2000) and BOSS that use a mosaic CCD camera (Gunn et al. 1998) plus multi- object spectrograph on a dedicated 2.5 m telescope (Gunn et al. 2006) at Apache Point Observatory. Between 2000–2008, the SDSS I/II completed spectroscopy over 9380 deg2 and obtained over 1.6 million spectra in total, including more than 105000 quasars (e.g., Richards et al. 2002; Abazajian et al. 2009; Schneider et al. 2010). BOSS is observing a sample of 210000 quasars, the majority of which are at z > 2.2 (Ross et al. 2012; ≈ Pˆaris et al. 2012), with the main scientific motivation being to measure the baryon acoustic oscillation (BAO) feature in the Lyman-α forest (e.g., Busca et al. 2013; Slosar et al. 2013). Using an improved spectrograph, BOSS spectra have coverage between 3600–10000 A˚ at a resolution of 1300–3000 (see Dawson et al. 2013; Smee et al. 2013). In addition to its primary quasar program, BOSS is executing several ancillary projects (see Dawson et al. 2013) including one focused on investigating the dynamics of quasar winds over multi-year timescales. This project re-observes selected bright BAL quasars that have previous SDSS spectral observations to enable a high-quality and rel- atively unbiased study of BAL variability over multi-year timescales in the rest frame of the quasar. The project targets were selected using information from the catalog of BAL quasars for SDSS DR5 (Gibson et al. 2009) and the SDSS DR5 quasar catalog (Schneider et al. 2007). The details of the BAL target selection are described in Section 2.1 of Filiz Ak et al. (2012). Briefly, the 2005 selected targets are optically bright (i < 19.3) BAL quasars with redshifts 0.48 z 4.65. The observed SDSS spectra of these targets have ≤ ≤ a signal-to-noise ratio per 0.4 A˚ pixel at 1650–1750 A˚ of SN1700 > 6; SN1700 is defined for the continuum redward of C iv and should not be affected by BAL absorption. The targets −1 were chosen to have a modified balnicity index BI0 > 100 km s . BI0 is defined by Gibson et al. (2008) using the following equation: −25000 f(v) BI0 1 Cdv. (3.1) ≡ Z0 − 0.9 where f(v) is the normalized flux density as a function of velocity, v, and C is a constant which is equal to 1.0 only when a trough is wider than 2000 km s−1, it is otherwise 0.0. In this study, we use SDSS spectra observed between MJD 51602 and 54557 (2000 February 28 to 2008 January 04) and BOSS spectra observed between MJD 55176 and 56109 (2009 December 11 to 2012 July 01); i.e., we utilize spectra taken after the completion of hardware commissioning for both SDSS and BOSS. Between these dates, 1087 of the 2005 targets from the ancillary project were observed by BOSS. 46 3.2.2 Sample Selection In this section we present the selection criteria used to create a “main sample” for studying C iv and Si iv BAL variability on multi-year timescales in the rest frame. We select our main-sample BAL quasars from the targets observed via the ancillary project based on the following criteria: 1. To enable more robust continuum fits, we select only the quasars that have spectral coverage of the relatively line free (RLF, see Section 3.3.1) windows blueward of the Si iv line as well as redward of the C iv line; these windows play a key role in constraining the fitted continuum. Thus, we utilize the quasars that have z > 2 from the sample of 1087 observed objects. (619 quasars) 2. The targeted sample of quasars was required to have SN1700 > 6 for the SDSS spec- trum, although higher values of SN1700 are advantageous for the study of moderate or weak BAL variations. We have thus chosen to utilize quasars that have SN1700 > 10 for both the SDSS and BOSS spectra; visual inspection shows that this choice pro- vides a good balance between high spectral quality and large sample size. (356 quasars out of 619) 3. As in Filiz Ak et al. (2012), to avoid confusion between emission-line and BAL vari- ability, we consider only the BAL troughs that are significantly detached from the C iv and Si iv emission lines by setting velocity limits for BAL troughs. We consider the BAL regions of each transition that lie between 3000 and 30000 km s−1. To − − select the quasars with moderate-to-strong BAL troughs, we require that the bal- nicity index of both the SDSS and BOSS spectra of the main-sample quasars have BI′ > 100 km s−1; a consistent threshold for both SDSS and BOSS is required to avoid biases in our later analyses. BI′ is defined by Filiz Ak et al. (2012) using the following equation: −30000 f(v) BI′ 1 Cdv. (3.2) ≡ Z−3000 − 0.9 Similar to the BI0 definition, in this equation f(v) is the normalized flux density as a function of velocity, v, and C is a constant which is equal to 1.0 only when a trough is wider than 2000 km s−1, it is otherwise 0.0. (297 quasars out of 356) 4. We have rejected six quasars (2%) from our main sample because of difficulties in defining the continua and/or emission lines in their spectra. Spectra of these quasars possess strong absorption lines of many transitions causing large systematic uncer- tainties in BAL measurements. (291 quasars out of 297) Based on these criteria, we selected 291 BAL quasars as our main sample. All 291 quasars have at least one observation from SDSS and one from BOSS, and 22% have 47 additional SDSS and/or BOSS observations. In total, our sample contains 699 spectra of main-sample quasars that cover rest-frame timescales from 5.9 hr to 3.7 yr. We have cross-matched our 291 main-sample quasars with the catalog of quasar properties from SDSS DR7 (Shen et al. 2011) and obtained their absolute i-band magni- tudes, Mi, estimated bolometric luminosities, LBol, Eddington-luminosity ratios, LBol/LEdd, and virial black-hole mass estimates, MBH. In addition we use the radio-loudness parame- ter, R, defined as R = f /f , where f is the radio flux density at rest-frame 6 cm 6cm 2500A˚ 6cm and f is the optical flux density at rest-frame 2500 A.˚ Shen et al. (2011) calculated 2500A˚ the R parameter using radio emission detected in Very Large Array (VLA) Faint Images of the Radio Sky at Twenty-Centimeters (FIRST; Becker et al. 1995) observations. We obtain redshift values from Hewett & Wild (2010), and these are used throughout. We illustrate some of the basic properties of our sample in Figures 3.1 and 3.2. Figure 3.1 shows Mi vs. redshift for the main sample of this paper, the 2005 targets of the BOSS ancillary project on BAL quasars, and all SDSS DR5 BAL quasars. The Mi distribution of the main-sample quasars spans about the same range as that of the general ancillary program targets from z = 2–4, covering a factor of 5 in luminosity at any ≈ given redshift. Figure 3.2 compares the i-band apparent magnitude distributions of our main sample, the 2005 BOSS ancillary project targets, and the BAL quasars identified in the SDSS DR10 quasar catalog (Pˆaris et al. 2013). It is clear that the main sample and ancillary project effectively cover the brightest BAL quasars that generally provide the highest quality SDSS and BOSS spectra. The mean i-band magnitude is 18.4 for the main-sample quasars and 18.7 for the ancillary project targets. Using the catalog of BAL quasars for SDSS DR5 (Gibson et al. 2009), we found that 22 of the 291 main- sample quasars possess Al iii BALs and thus are identified as low-ionization BAL quasars. Additional unidentified low-ionization BAL quasars may be present; e.g., among objects that lack spectral coverage of the important Mg ii low-ionization transition. 3.3 Data Preparation and Analysis 3.3.1 Basic Spectral Preparation For the purpose of investigating BAL variability, we compared the multi-epoch spectral observations of our main-sample quasars. We have normalized each spectrum by a model for the continuum following the procedure in Section 3.1 of Filiz Ak et al. (2012). Briefly, we first correct the spectra for Galactic extinction using the AV values from Schlegel et al. (1998) and then transform from the observed frame to the rest frame using the redshift values from Hewett & Wild (2010). We remove the pixels from the spectra that contain significant night-sky line residuals that are flagged by the SDSS and BOSS data-reduction pipelines. To reconstruct the underlying continuum, we define RLF windows to be the following spectral regions: 1250–1350 A˚ , 1700–1800 A˚ , 1950–2200 A˚ , 2650–2910 A˚ , 3950–4050 A.˚ 48 −30 −29 −28 i −27 M −26 −25 Main Sample −24 Targets DR5 BAL Quasars −23 1 2 3 4 5 z Fig. 3.1: Absolute i-band magnitude, Mi, vs. redshift for the main sample of this paper (blue squares), the 2005 targets of the BOSS ancillary project on BAL quasars (open circles), and all SDSS DR5 BAL quasars (dots). We fit the RLF windows of each spectrum with an intrinsically reddened power-law con- tinuum model where we use Small-Magellanic-Cloud type reddening. To exclude the data points that deviate from the fit by more than 3σ, we apply an iterative sigma-clipping algorithm using a non-linear least squares fit. We calculate the continuum uncertainties using ∆χ2 confidence-region estimation for 68.3% confidence bounds. Throughout this work, we propagate the continuum uncertainties into the uncertainties on rest-frame EW measurements. As in previous studies (e.g., Lundgren et al. 2007; Filiz Ak et al. 2012), we do not model the emission lines since investigation of emission-line characteristics is beyond the scope of this study (also see Section 3.2.2). 3.3.2 Identification and Measurements of BAL Troughs As is common practice (e.g., Trump et al. 2006; Gibson et al. 2008, 2009; Allen et al. 2011; Filiz Ak et al. 2012), we smoothed each spectrum using a Savitzky-Golay algorithm to perform local linear regression for three consecutive data points (see Section 3.2 of Filiz Ak et al. 2012). We utilize normalized and smoothed spectra only for identification of BAL troughs; unsmoothed spectra are used for further calculations. Only BAL troughs in the velocity range 3000 to 30000 km s−1 are considered (see Section 3.2.2); the small fraction − − 49 1000 100 10 Number of Quasars 1 Main Sample Targets DR10 BAL Quasars 0 16 17 18 19 20 21 22 i−band magnitude Fig. 3.2: Comparison of the i-band apparent magnitude distributions of the main sample of this paper (blue), the 2005 targets of the BOSS ancillary project on BAL quasars (gray), and the BAL quasars identified in the SDSS DR10 quasar catalog (dashed red line). Note that we are targeting the brightest BAL quasars in the SDSS sky area in order to obtain spectra of the highest possible quality. ( 2.5%) of extremely high-velocity C iv BAL troughs exceeding the 30000 km s−1 limit ≈ − were removed by considering Si iv BAL troughs at corresponding velocities. The canonical definition of BAL troughs (see Equation 3.2) was developed for the purpose of finding BAL troughs in a single-epoch spectrum. However, our primary purpose here is investigating BAL variability in multi-epoch spectra. Therefore, we adopt a modified BAL-trough definition more appropriate for our purpose that is strongly motivated by the canonical definition and reduces to it for single-epoch data. Our adopted BAL-trough definition utilizes the information from all available spectral observations of a quasar. As is well known, absorption troughs are sometimes isolated and sometimes appear in complexes in which single troughs may split or adjacent troughs may merge over time. To address these complications, we treat each BAL complex as a single BAL trough. We identify BAL complexes using the following algorithm (see Figure 3.3): 50 1. We first identify BAL and mini-BAL troughs (hereafter just “troughs”) under the canonical definition in each single-epoch spectrum of a quasar.1 We set the maximum velocity of a trough to be vmax,t and the minimum velocity to be vmin,t. 2. We select vmin,t of the highest velocity trough in the first-epoch spectrum and compare the corresponding velocities in all the available spectra. If the vmin,t velocity intersects any trough region in the other available epochs, we re-assign vmin,t to be the lowest velocity of this intersecting trough and repeat the comparison to the other available spectra. If the vmin,t velocity does not intersect any trough region in the other available spectra, we set vmin,t to be the minimum red-edge velocity of the BAL complex, vmin. 3. To define the maximum velocity of the complex, we take the vmax,t of the lowest- velocity trough associated with the complex and compare with the other available epochs. If the vmax,t velocity in the other spectra intersects a trough, we set vmax,t to be the highest velocity of this intersecting trough. If the vmax,t velocity does not intersect any trough region in the other available spectra, we set vmax,t to be the maximum blue-edge velocity of the BAL complex, vmax. In this algorithm each trough can be associated with only one trough complex. Each trough complex includes at least one trough which is wider than 2000 kms−1 lying between 3000 − and 30000 km s−1. − After implementing the above algorithm, we define each BAL complex lying between vmax and vmin as a distinct BAL trough. Here, vmax is the maximum velocity taken to be the velocity at the blue edge for any trough associated with the complex across all available spectra of each quasar. Similarly, vmin is defined using the red-edge velocities. After an automated identification of BAL troughs using our algorithm above, we visually inspect all the available spectra of each main-sample quasar. The inspection shows that our approach for BAL-trough identification is appropriately implemented for both C iv and Si iv BAL troughs. We found that, in the majority of cases, the complex would be identified as a single BAL trough under the canonical definition in at least one of our epochs. Moreover, our adopted BAL-trough definition produces the same results as the canonical definition for non-merging and non-splitting BAL troughs that lie between constant vmax and vmin in all the available spectral observations. Using our adopted BAL-trough definition we identified a total of 428 distinct C iv and 235 distinct Si iv BAL troughs in the 699 main-sample spectra. Figure 3.3 illustrates our adopted BAL-trough definition and the canonical one using all the available spectra of the quasar SDSS J090944.05+363406.7. If we apply the canonical BAL-trough definition in this example, the two adjacent BAL troughs seen in 1Mini-BALs are defined using Equation 3.2 but for a trough width of 500–2000 kms−1 (cf. Hamann & Sabra 2004). 51 the last epoch would be identified as two distinct BALs. However, the same structure appears as one distinct BAL trough in the second-epoch spectrum and in the first-epoch spectrum appears as two mini-BALs along with one BAL trough. As is clearly seen in this example, our adopted BAL-trough definition produces more physically meaningful results than the canonical definition for the purpose of studying variability in multi-epoch spectra. Moreover, we select the vmax and vmin velocities using information from all the available spectra instead of only a single-epoch observation. If we were to take the vmax and vmin velocities as the blue-edge and red-edge velocities of the complex where all absorption has merged to one BAL trough (i.e., the second-epoch spectrum in this example), we would lose the pertinent information from the part of the BAL troughs which extend beyond these velocities in the other-epoch spectra (i.e., the first-epoch spectrum in this example). In this study, we investigate BAL-trough variability on a large range of rest-frame timescales. The spectral observations from SDSS-I/II and BOSS provide coverage of long timescales, typically longer than 1 yr. To sample shorter timescales, we use the additional observations from SDSS and/or BOSS that are available for more than 20% of our main- sample quasars. To avoid the repeat examination of BAL troughs and the associated multi-counting biases, we utilize only the two-epoch spectra for each quasar that give 2 the minimum sampled rest-frame timescale, ∆tmin. Thus, all of our calculations below consider only two-epoch observations of each BAL trough. By selecting ∆tmin, we sample rest-frame timescales from a few hours to a few years. Figure 3.4 shows the distribution of minimum sampled rest-frame timescales, ∆tmin, for distinct BAL troughs identified in our main-sample spectra. The ∆tmin values range between 5.9 hr and 3.7 yr with a median of 2.1 yr. Given that C iv BAL troughs are not always accompanied by Si iv BAL troughs, we ran a two-sample Kolmogorov-Smirnov (KS) test to compare the sampled timescale distributions of C iv and Si iv BAL troughs and found no significant difference. In some sections of this study, we focus on BAL-trough variability characteristics solely on multi-year timescales. Therefore, we define another timescale of ∆tmin,1 to sample minimum rest-frame timescales of more than 1 yr. By this definition, we select the same non-repeating sample of distinct BAL quasars as with the ∆tmin selection. ∆tmin,1 ranges between 1–3.7 yr with a median of 2.3 yr. Compared to previous studies (e.g., Lundgren et al. 2007; Gibson et al. 2008, 2009; Capellupo et al. 2011, 2012), and especially those focusing on multi-year timescales, we have a significantly larger (by about an order-of- magnitude) BAL-trough sample. We measure the rest-frame EW of each BAL trough in each epoch and calculate the uncertainties on EWs using Equations 1 and 2 of Kaspi et al. (2002), where uncertainties 2For example, if a quasar has observations that sample rest-frame timescales of 0.01, 0.02, 0.03, 2.00, 2.01, and 2.03 yr, we select only the two-epoch spectra that sample the 0.01 yr timescale. Note that the 2.00, 2.01, and 2.03 yr timescales are nearly the same (agreeing to within 1.5%) and provide little independent information. Thus, using all three of these timescales would result in this object being inordinately weighted in statistical characterizations of BAL variability on 2 yr timescales (i.e., causing multi-counting bias). ≈ 52 v v max min λ MJD 52703 MJD 55623 Normalized Flux Density F MJD 55946 C IV −30000 −25000 −20000 −15000 −10000 −5000 0 v (km s−1) Fig. 3.3: Example of our adopted BAL-trough definition illustrated using the three available spectra of the quasar SDSS J090944.05+363406.7. The three normalized spectra for this quasar are arbitrarily offset in flux for clarity of presentation. Horizontal dotted lines show the continuum levels for each spectrum, and the tick marks on the y-axis show the zero level for each spectrum. Horizontal black bars show absorption lines with ∆v 2000 km s−1, ≥ corresponding to the traditional BAL definition, and horizontal gray bars show absorption lines with ∆v = 500–2000 km s−1. The variable absorption complex is complicated; in some spectra only part of it is classified as a BAL under the canonical definition. We use all three available spectra to define the minimum and maximum velocities of the trough. The horizontal double green bar shows the resulting BAL trough lying between vmin and vmax (see Section 3.3.2). 53 75 C IV 428 BAL Troughs 50 25 ∆t (yr) Si IV 40 235 BAL Troughs Number of BAL Troughs 20 0.0 0.25 0.5 0.75 1.0 1.25 1.5 1.75 2.0 2.25 2.5 2.75 3.0 3.25 3.5 3.75 ∆t (years) min Fig. 3.4: Distributions of the minimum sampled rest-frame timescale, ∆tmin, for 428 distinct C iv (upper panel) and 235 distinct Si iv (lower panel) BAL troughs present in the spectra of the main-sample quasars. We have significant trough statistics on timescales as long as 3–3.5 yr. 54 are derived by propagating the continuum-estimation errors (see Section 3.3.1) and the observational errors of each contributing pixel. In addition, we measure the weighted centroid velocity, vcent, for each BAL trough; i.e., the mean of the velocities where each data point is weighted with its distance from the normalized continuum level. We also calculate an average BAL-trough depth, dBAL, which is the mean distance from the normalized continuum level for each data point of a BAL trough. 3.3.3 Identification and Measurements of Variable BAL Troughs As one approach to identify variable BAL troughs in our main sample, we select BAL troughs showing significant EW variations. To calculate EW variations, ∆EW, and uncertainties on this quantity, σ∆EW, we use the following equations: 2 2 ∆EW = EW2 EW1, σ∆EW = σEW2 + σEW1 (3.3) − q where EW1 and EW2 are the EWs measured from two-epoch spectra that are observed iv iv at times t1 and t2. In our sample, the mean σ∆EW is 0.5 A˚ for C and 0.4 A˚ for Si ∆EW BAL troughs. Similarly, we calculate fractional EW variations, hEWi , and corresponding uncertainties, σ ∆EW , with the following equations: hEWi ∆EW (EW EW ) = 2 − 1 , EW (EW + EW ) 0.5 h i 2 1 × 4 (EW2σEW1 + EW1σEW2 ) σ ∆EW = × (3.4) EW 2 h i (EW2 + EW1) We identify 248 variable C iv BAL troughs and 119 variable Si iv BAL troughs showing EW variations at a significance level of more than 3σ on timescales of more than iv iv 1 yr (∆tmin,1). Similarly, we identify 223 variable C BAL troughs and 99 variable Si BAL troughs by comparing the two-epoch spectra that sample the ∆tmin timescales in the rest frame. Considering that variations tend to occur in portions of BAL troughs (e.g., Gibson et al. 2008; Capellupo et al. 2011), as an alternative approach, we define a variable BAL trough to have at least one variable region. For the purpose of determining regions in each BAL trough where a variation has occurred, we compare two-epoch spectra of each quasar. Since a proper comparison requires consideration of the signal-to-noise ratio of each spectrum, we define a measurement of the deviation between two observations for each pixel in units of σ using the following equation: f2 f1 Nσ(λ)= − (3.5) 2 2 σ2 + σ1 q 55 where f1 and f2 are the normalized flux densities and σ1 and σ2 are the normalized flux- density standard deviations at wavelength λ. Both σ1 and σ2 include observational errors and uncertainties on the estimated continuum model. Similarly to Gibson et al. (2008), we identify variable regions of BAL troughs to be where an absorption feature is detected with Nσ 1 or Nσ 1 for at least five consecutive data points. This requirement ≥ ≤ − 1 allows detection of variable regions wider than 275 km s− . Selection using a smaller ≈ number of consecutive data points as the requirement may cause non-physical observational errors to be indistinguishable from the variable regions of BAL troughs. On the other hand, requiring a larger number of consecutive data points will cause non-detection of narrow variable regions. By requiring the number of data points to be 5, we require the ≥ significance of variations to be >99.9%. We have identified 903 variable regions for C iv BAL troughs and 294 variable regions iv for Si BAL troughs for variations on timescales of more than 1 yr (∆tmin,1). We also identified 757 variable regions for C iv BAL troughs and 232 variable regions for Si iv BAL troughs by comparing the two-epoch spectra that sample the ∆tmin timescales in the rest frame. The number of BAL troughs having at least one variable region is 294 for C iv and 119 for Si iv on timescales of more than 1 yr. Comparing the two approaches to variable BAL-trough identification, we found that 26 C iv BAL troughs showing EW variations at more than 3σ significance do not have a variable region satisfying our requirements, although several narrow variable regions in these BAL troughs collectively produce EW variations at more than 3σ significance. We also found that 72 C iv BAL troughs having one variable region do not show EW variations at more than 3σ significance. In these 72 cases, a narrow variable region in a strong BAL trough cannot produce an EW variation at more than 3σ significance due to statistical dilution by the rest of the trough. Similarly, we found that 29 Si iv BAL troughs showing EW variations at more than 3σ significance do not have a variable region satisfying our requirements, and 29 Si iv BAL troughs having one variable region do not show EW variation at more than 3σ significance. We will refer to both of these approaches to variable BAL trough identification in the following sections. We present our measurements for C iv and Si iv BAL troughs in Tables ?? and ?? and for C iv and Si iv variable regions in Tables 3.4 and 3.5, respectively. 3.4 Results on BAL Variability In this section, we present the results of our BAL variability investigations utilizing the multi-epoch observations of 428 distinct C iv and 235 distinct Si iv BAL troughs in the 699 main-sample spectra of 291 quasars. We examine the fraction of variable BAL troughs and BAL quasars (Section 3.4.1), the velocity widths of the variable regions of BAL troughs (Section 3.4.2), EW variations as a function of timescales (Section 3.4.3), the distribution of EW variations (Section 3.4.4), EW variations as a function of BAL profile properties (Section 3.4.5), relative EW variations between C iv and Si iv BAL troughs (Section 3.4.6), 56 correlated EW variations in BAL quasars with multiple troughs (Section 3.4.7), and EW variations as a function of quasar properties (Section 3.4.8). 3.4.1 Fraction of BAL Troughs and BAL Quasars Showing Variability We calculate the fraction of BAL troughs showing variability and the fraction of quasars showing BAL-trough variability in our main sample considering the two different variability identification approaches explained in Section 3.3.3. First, requiring a variable BAL trough to show an EW variation at more than 3σ significance, we find that the +3.9 iv +5.1 iv fraction of variable BAL troughs is 57.9−3.7% for C and 50.6−4.6% for Si on timescales of 1–3.7 yr. Figure 3.5 presents the cumulative fraction of variable BAL troughs; the y-axis shows the cumulative fraction of BAL troughs with EW variations of more than a given threshold ∆EW . Although the cumulative fraction of variable BAL troughs decreases for | | large ∆EW , it remains significant even for threshold ∆EW values as large as 5 A.˚ | | | | 100 100 90 C IV 90 Si IV 80 80 70 70 60 60 50 50 40 40 30 30 Cumulative Fraction of Variable BAL Troughs (%) 20 20 10 10 0 0 0 5 10 15 0 5 10 15 Threshold |∆EW|(A)˚ Fig. 3.5: Cumulative fraction of BAL troughs with a given threshold ∆EW for C iv and | | Si iv troughs. The open histograms show the cumulative fraction of BAL troughs with EW variations of more than a given threshold. The shaded parts of the histograms indicate the fraction of BAL troughs showing EW variations of more than 3σ significance on timescales of 1–3.7 yr. We also compare the fraction of variable C iv and Si iv BAL troughs from the same absorbing material. Given that the Si iv region between 13000 and 30000 km s−1 − − can have contamination by emission and/or absorption lines such as C ii (1335 A),˚ O i (1306 A),˚ and Lyα+N v, we select a sample of 136 C iv BAL troughs lying between 3000 − 57 and 13000 km s−1 that are accompanied by Si iv BAL troughs at corresponding velocities. − In this sample, we found that the fraction of BAL troughs showing EW variations at more +6.8 iv +6.3 iv than 3σ significance is 50.0−6.0% for C and 53.7−5.4% for Si on timescales of 1–3.7 yr. These fractions indicate that C iv and Si iv BAL troughs at corresponding velocities are about equally likely to vary. In addition to considering the fraction of variable BAL troughs, we also calculate the fraction of quasars showing BAL-trough variability. Requiring a variable BAL trough to show an EW variation at more than 3σ significance, we found the fraction of quasars iv +4.9 showing C BAL variations to be 62.2−4.6% (181/291). Considering a total of 181 quasars showing Si iv BAL absorption, we found the fraction of quasars showing Si iv BAL varia- +6.3 tions to be 59.1−5.7% (107/181). Using multi-epoch observations of 24 quasars, Capellupo et al. (2011) found that the fraction of quasars showing C iv absorption variations is 39% on timescales of 0.35–0.75 yr and 65% on timescales of 3.8–7.7 yr. Capellupo et al. (2012) reported that 11 out of 19 (58%) quasars exhibited Si iv absorption variations on timescales of 3.8–7.7 yr. Considering that our data sample timescales of 1–3.7 yr with an average of 2.3 yr, both of our results broadly show consistency with the results of the Capellupo et al. (2011) study. Alternatively, we consider variable BAL troughs to be those with at least one vari- able region detected in the trough (see Section 3.3.3). We find that the fraction of BAL +4.3 iv +5.1 iv troughs showing variability is 68.6−4.0% for C and 50.6−4.6% for Si on timescales of 1–3.7 yr. This approach is more sensitive to local variations in BAL troughs; a narrow variable region in a wide BAL trough may not produce an EW variation at more than 3σ significance. We further investigate the number of variable regions as a function of velocity, v, that is measured relative to the quasar redshift. Figure 3.6 presents the number of vari- able regions found at a particular velocity for C iv and Si iv BAL troughs for variations iv on timescales of more than 1 yr (∆tmin,1). C variable regions are found across a wide range of velocities, and the number of variable regions appears to peak in the range be- tween 9000 and 21000 km s−1 in concert with the number of BAL troughs. Figure 3.6 − − also displays the percentage of BAL regions showing variability for C iv and Si iv that is calculated from the ratio of the number of variable regions to the number of BAL troughs found at a particular velocity. We found that the percentage of C iv BAL regions show- ing variability is roughly constant at around 30–40% at velocities of 3000 km s−1 to 1 − 25000 km s− and rises in the few highest-velocity bins; we have verified that this trend − is statistically significant. Capellupo et al. (2011) also investigated the fraction of variations as a function of outflow velocity on short (0.35–0.75 yr) and long (3.8–7.7 yr) timescales. Consistent with our results, they found that C iv BAL troughs tend to be more variable at higher velocities. The Si iv region can have contamination by emission and/or absorption lines, and the superposition of these emission lines and Si iv BALs may prevent BAL troughs from 58 100 0.8 C IV 80 0.6 60 0.4 Percentage of BAL Regions 40 Showing Variability 20 0.2 0 0 0.8 40 Si IV 0.6 30 Number of Variable Region 20 0.4 10 0.2 0 0 −3 −2.8 −2.6 −2.4 −2.2 −2 −1.8 −1.6 −1.4 −1.2 −1 −0.8 −0.6 −0.4 v (104 km s−1) Fig. 3.6: Number of times a variable region is found at a particular velocity in C iv (upper panel) and Si iv (lower panel) BAL troughs for variations on timescales of more than 1 yr. In the upper panel, the dashed dark-blue line along with the right y-axis shows the percentage of BAL regions showing variability in C iv.C iv variable regions are found across a wide range of velocities. The Si iv region can have contamination by emission lines such as C ii (1335 A),˚ O i (1306 A),˚ and Lyα+N v, causing the apparent decrease in the number of variable Si iv regions between 13000 and 30000 km s−1; the histogram − − for Si iv should not be interpreted physically in this velocity range. Therefore, we do not show the percentage of variable regions for Si iv between 13000 and 30000 km s−1. − − 59 continuously lying at least 10% under the continuum level and thus may lead to non- detection. A visual inspection showed that in some cases BAL troughs appear to be broken into two or more narrow sections by such emission lines, causing the apparent absorption features to fail to satisfy the criteria to be identified as a BAL trough. Consequently, these effects also cause a decrease in the number of variable Si iv BAL troughs. Due to these effects, the number of variable Si iv BAL troughs and therefore variable Si iv regions decreases between 13000 and 30000 km s−1; the histogram for Si iv should not be − − interpreted physically in this velocity range. Thus, we show the percentage distribution of Si iv BAL troughs only for a region between 3000 and 13000 km s−1. − − 3.4.2 Velocity Widths of Variable Regions of BALs Gibson et al. (2008) showed that variations tend to occur only in portions of BAL troughs. To investigate the distribution of velocity widths of variable BAL regions, we iv calculate ∆vVR values for 903 variable regions in C BAL troughs and 294 variable regions iv in Si BAL troughs for variations on timescales of more than 1 yr, where ∆vVR is the velocity width of a variable region. Figure 3.7 shows the ∆vVR distributions of variable regions detected in C iv and Si iv BAL troughs. Consistent with Gibson et al. (2008), we find that the number of variable regions, in both C iv and Si iv, rises toward small velocity widths down to our velocity width measurement limit of 275 km s−1 (see Section 3.3.3). ≈ We found that the mean of the ∆vVR measurements for C iv and Si iv BAL troughs are 713.6 kms−1 and 592.8 kms−1, respectively. To examine the fraction of a BAL trough that is variable, we define f∆v as the sum of the velocity widths of all varying regions in a BAL trough divided by the BAL iv trough velocity width, ∆v. Figure 3.8 shows the f∆v distribution for C BAL troughs on iv timescales of 1–3.7 yr and the f∆v distribution as a function of ∆v for C . We find that iv iv iv the mean of f∆v is 0.20 for C and 0.13 for Si , indicating that Si variable regions on average tend to be narrower than C iv variable regions. Figure 3.8 also suggests that narrow C iv BAL troughs tend to have a larger fraction of variable regions compared to wide C iv BAL troughs. We also investigate the position of each variable region in a given BAL trough to assess if the incidence of variability depends upon the relative velocity within the trough. vcent−vmid We calculate the normalized relative velocity in the trough, vnrt. Here vnrt is |vcent−vmax| for vcent−vmid the blue part of the trough and |vcent−vmin| for the red part of the trough, where vmid is the mid velocity of a variable region. Figure 3.9 shows the number of variable regions found at a given vnrt. A non-parametric triples test (Randles et al. 1980) shows that the distribution of vnrt shows no significant evidence of asymmetry (P = 0.55), and the distribution is relatively constant across almost the entire width of a trough. Consistent with our results, Capellupo et al. (2011) found no evidence for a higher incidence of variability with positive or negative velocity offset, but we establish this result with substantially better statistics. 60 40 C IV 30 20 10 0 40 Si IV 30 Percentage of Variable Regions 20 10 0 500 1000 1500 2000 2500 3000 3500 ∆v (km s−1) VR Fig. 3.7: Percentage of variable regions with a given velocity width in C iv (upper panel) and Si iv (lower panel) BAL troughs for variations on timescales of more than 1 yr. We found seven variable regions for C iv and one variable region for Si iv with velocity widths of 4500–7500 km s−1 that we do not show in this figure. For both C iv and Si iv, the number of variable regions rises rapidly toward small velocity widths down to our velocity width measurement limit. 61 1 200 C IV C IV 0.8 150 0.6 v ∆ 100 f 0.4 Number of BAL Troughs 50 0.2 0 0 0 0.2 0.4 0.6 0.8 1 0 5000 10000 15000 20000 25000 f −1 ∆v ∆v (km s ) Fig. 3.8: Distribution of f∆v (left panel), and f∆v distribution as a function of BAL trough width ∆v (right panel). Vertical black bars in the right panel show the standard deviation −1 around the mean for non-zero f∆v values in given 100 km s wide ∆v bins. Both of the panels are for C iv BAL troughs on timescales of 1–3.7 yr. 3.4.3 EW Variations as a Function of Timescale Previous BAL-variability studies (e.g., Gibson et al. 2008, 2010; Capellupo et al. 2011) have found that C iv BAL-trough variability is larger for longer timescales. This is expected since, e.g., quasar variability in general is larger on longer timescales (e.g., Vanden Berk et al. 2004). In order to investigate EW variations as a function of timescale with a larger sample over a wide range of rest-frame timescales, we utilize the ∆EW and ∆EW/ EW values of each distinct BAL trough from the two-epoch spectra for timescales h i of ∆tmin (see Equations 3.3 and 3.4). In Figure 3.10, we show EW variations for C iv and Si iv BAL troughs as functions of ∆tmin. For comparison, we also include the data from Barlow (1993) that correspond to a timescale range mainly between 0.2–1 yr and from Gibson et al. (2008) that extend the timescales up to 6.1 yr. To display the spread of ∆EW, we calculate the standard deviation of our data using a sliding window containing 20 time-ordered data points; we statistically remove the mean EW error in each window from the standard deviation (via standard error propagation). The curves of standard deviation indicate an increase of EW variations with increasing rest-frame timescale both for C iv and Si iv BAL troughs. This trend is consistent with that of previous BAL-variability studies. The majority of the Barlow (1993) data lie between the standard-deviation curves that are calculated from our data. Although the data from Gibson et al. (2008) sample 62 60 C IV 50 40 30 20 Number of Variable Regions 10 0 −1 −0.5 0 0.5 1 v (km s−1) nrt Fig. 3.9: The number of variable regions found at a given vnrt. The distribution is relatively constant across the entire trough width. longer timescales than our data, the trend of the standard-deviation curves shows general agreement regarding the increase of EW variations with increasing timescale. Figure 3.11 shows fractional EW variations for C iv and Si iv BAL troughs against ∆tmin. The curves of standard deviation are illustrated both for C iv and Si iv BAL troughs. The spread of the curves indicates an increase of fractional EW variations with increasing rest-frame timescales. As for Figure 3.10, we include the data from Barlow (1993) and Gibson et al. (2008) for comparison purposes. Figures 3.10 and 3.11 demonstrate that our data points are consistent with zero variation for timescales ∆tmin < 0.01 yr (i.e., 3.6 days) in the rest frame given the mea- surement errors. This result is expected given the fact that troughs are unlikely to show significant variations on timescales of hours or days. Therefore, any observed variations on these timescales provides an empirical estimate of the total systematic and measurement errors of our data and data-processing methods. For a more quantitative assessment, we calculated the median deviation in units of σ for ∆t < 0.01 yr and found that it is 1 min ≈ for both C iv and Si iv. To quantify the relationship between EW variations and timescale, we calculate the mean of ∆EW and ∆t for bins containing 15 time-ordered data points. From a robust | | min linear-regression model using the bisquare weight function (Press et al. 1992), we found a fit of log ∆EW = (0.258 0.031) log ∆t + (0.289 0.027) (3.6) | C IV| ± × min ± 63 15 C IV 10 5 0 −5 This study −10 Barlow (1993) −15 Gibson et al. (2008) ˚ A) ( Si IV EW ∆ 5 0 −5 0.001 0.01 0.1 1 2 3 4 56 ∆t (yr) min Fig. 3.10: EW variation, ∆EW, vs. the minimum sampled rest-frame timescale, ∆tmin, for C iv (upper panel) and Si iv (lower panel) BAL troughs. The data are from this study (light blue circles), Barlow (1993; red triangles), and Gibson et al. (2008; dark blue squares). The red solid curves indicate the standard deviation derived from the data in this study, calculated using a sliding window containing 20 time-ordered data points. The standard deviation of ∆EW increases with increasing rest-frame timescale. 64 2 C IV 1 0 −1 This study Barlow (1993) −2 Gibson et al. (1998) 2 0.001 0.01 0.1 1 2 3 4 56 Si IV ∆t (yr) EW / 0 −1 −2 0.001 0.01 0.1 1 2 3 4 56 ∆t (yr) min Fig. 3.11: Same as Figure 5 but for fractional EW variation, ∆EW/ EW . h i 65 where the units of ∆EW and ∆t are A˚ and yr, respectively. Similarly, for fractional | | min EW variations we found a fit of ∆EW log = (0.283 0.042) log ∆t + ( 0.737 0.037). (3.7) EW ± × min − ± C IV h i In Figure 3.12, we display the fraction of C iv BAL troughs showing variability at more than 3σ significance as a function of timescale. We calculate the mean timescale and the fraction of variable BAL troughs for 20 time-ordered data points for variations on timescales of less than 1 yr, and for two equal-size bins for variations of more than 1 yr. Consistent with Gibson et al. (2010) and Capellupo et al. (2013), our results indicate the incidence of variability increases with time. 0.7 0.6 0.5 0.4 0.3 0.2 0.1 Fraction of Variable BAL Troughs 0 0.001 0.01 0.1 1 2 3 ∆t (yr) min Fig. 3.12: The fraction of C iv BAL troughs showing variability at more than 3σ significance as a function of timescale. We calculate the counting errors following Gehrels (1986). We also investigate the rate-of-change of EW variations on short and long timescales by calculating ∆EW /∆t for C iv and Si iv BAL troughs on timescales of ∆t = 0.01 1 yr | | min − and ∆t > 1 yr. Figure 3.13 displays the distribution of ∆EW /∆t for C iv and min | | Si iv BAL troughs on both short and long timescales. It also presents the distribution of ∆EW /∆t for BAL troughs with 3σ significance variations of EW in each panel. The | | ≥ distributions for variations with 3σ significance show that the average rate-of-change of ≥ EW variations is larger on short timescales. Gibson et al. (2010) found a similar result from the comparison of variations of BAL troughs for eight individual sources on short and 66 long timescales (see their Figure 9). Such behavior would be expected, for example, if BAL EWs execute a simple random walk with a step timescale of . 1 yr (see Section 3.5.4). On timescales longer than the step timescale then the observed ∆EW /∆t 1/√n, where n | | ∝ is the number of steps during the observed period. 3.4.4 Distribution of EW Variations In this section, we investigate several characteristics of BAL EW variability distribu- tions. We first examine the symmetry of distributions of BAL EW variations to assess and constrain any differences between the formation and decay timescales of BAL troughs. Fig- ure 3.14 presents the distributions of 428 C iv and 235 Si iv BAL-trough EW variations, ∆EW, and fractional EW variations, ∆EW/ EW , for variations on timescales of more h i than 1 yr. To maintain an equilibrium of BAL troughs in quasar spectra, we expect the weakening and strengthening of BAL troughs in a large sample to be balanced. To assess whether BAL-trough variations are balanced, we examine the mean values of the ∆EW and ∆EW/ EW distributions. We find that the mean of the ∆EW distribution for C iv h i BAL troughs is 0.082 0.184 A˚ and for Si iv BAL troughs is 0.101 0.163 A.˚ The error on − ± ± the mean is calculated following σ/√N where N is the number of BAL troughs. Similarly, we find the mean of the ∆EW/ EW distribution for C iv BAL troughs is 0.032 0.018 h i − ± and for Si iv BAL troughs is 0.003 0.029. These results indicate that the mean values − ± of the ∆EW and ∆EW/ EW distributions for C iv and Si iv BAL troughs are broadly h i consistent with zero. We next investigated if the ∆EW and ∆EW/ EW distributions are symmetric. An h i asymmetric distribution could be seen, for instance, if the strengthening and weakening of BAL troughs occur at different rates. For example, if a typical BAL trough forms rapidly and decays slowly, the distribution of BAL variations would be skewed to negative ∆EW (i.e., the number of weakening BAL troughs would be larger at any given time). Gibson et al. (2010) showed that the distribution of ∆EW for 23 C iv BAL troughs is reasonably symmetric. As can be seen in Figure 3.14, the ∆EW and ∆EW/ EW distributions do not h i show clear evidence of asymmetry. For a more quantitative examination, we use a non- parametric triples test. The results of this test on the ∆EW distributions for C iv and Si iv BAL troughs show no significant evidence of asymmetry (P = 0.45 and P = 0.40, respectively). Similarly, we found no significant evidence of asymmetry for the ∆EW/ EW h i distributions for C iv and Si iv BAL troughs (P = 0.25 and P = 0.04, respectively). As another approach to assess asymmetry, we compare the distributions of strength- ening and weakening BAL troughs by running a two-sample KS test for the distributions of BAL troughs with increasing and decreasing EWs. The test results show that the positive and negative parts of the ∆EW distributions for C iv and Si iv BAL troughs do not show significant inconsistency (P = 0.58 and P = 0.83, respectively). Similarly, we find that 67 70 ∆t = 0.01 − 1 yr ∆t > 1 yr min min 4 60 50 40 2 30 20 10 C IV 3 40 30 Number of BAL Troughs 2 20 1 10 Si IV 0 0 −2 −1 0 1 2 3 −2 −1 0 1 2 3 ˚ −1 log( |∆EW | / ∆t) ( A yr ) Fig. 3.13: Distributions of rate-of-change of EW , ∆EW /∆t, for C iv (upper panels) | | | | and Si iv (lower panels) BAL troughs both on short (left panels) and long (right pan- els) timescales. The shaded parts of the histograms show the BAL troughs with 3σ ≥ significance variations of EW. 68 C IV 100 Disappearing 10 1 100 Si IV Number of BAL Troughs 10 1 −10 0 10 −1 0 1 ˚ ∆ EW (A) ∆ EW / Fig. 3.14: Distributions of BAL-trough EW variations, ∆EW, and fractional EW vari- ations, ∆EW/ EW , for C iv (upper panels) and Si iv (lower panels) for variations on h i timescales of more than 1 yr. Black dashed lines show the best Gaussian fits to the data. Our main sample includes three of the disappearing C iv BAL troughs described in Filiz Ak et al. (2012); these are plotted with a dark gray histogram in the upper panels (see Section 3.5.3 for further discussion). Our main sample also includes one additional new case of C iv BAL disappearance that satisfies the disappearance criteria used in Filiz Ak et al. (2012); this case is plotted in the upper panel at ∆EW/ EW = 1.5 and is found h i − in the quasar SDSS J095901.24+550408.2. The cases of BAL disappearance appear to be an extreme of the overall distribution of BAL variability, rather than a distinct population of variability events. None of the disappearing C iv BAL troughs has a Si iv BAL trough at corresponding velocities at our observed epochs. 69 the positive and negative parts of the ∆EW/ EW distributions for C iv and Si iv BAL h i troughs do not show significant inconsistency (P = 0.12 and P = 0.60, respectively). Given that the test results show no significant evidence of asymmetry, we calculate the skewness of the ∆EW and ∆EW/ EW distributions to establish a basic limit upon h i the amount of asymmetry allowed by our data. We found that the skewness of the ∆EW distribution is 0.17 0.12 for C iv BAL troughs and 0.17 0.16 for Si iv BAL troughs. − ± − ± For the calculation of the error on skewness, we follow the standard approximate 6/N formula (Press et al. 1992). The skewness of the ∆EW/ EW distribution is 0.33 p0.12 h i − ± for C iv BAL troughs and 0.32 0.16 for Si iv BAL troughs. − ± We next investigated whether ∆EW and ∆EW/ EW are Gaussian distributed; fol- h i lowing the central-limit theorem, many random variability processes will generate Gaussian distributions. Although their sample of BAL troughs was not large enough to draw firm conclusions about the characteristics of BAL variation distributions, Gibson et al. (2010) suggested that ∆EW may not be normally distributed over time spans of months-to-years. We analyze the shape of the ∆EW and ∆EW/ EW distributions for C iv and Si iv BAL h i troughs considering variations longer than 1 yr using the Lilliefors normality test (Lilliefors 1967). The test shows that both the ∆EW and ∆EW/ EW distributions for C iv and h i Si iv BAL troughs are non-Gaussian (at a significance level of > 99.9%). Figure 3.14 shows the best Gaussian model distributions for comparison purposes in each panel. Since the mean values of the ∆EW distributions are consistent with zero, we calculate the best Gaussian models using µ = 0 and the measured standard deviation of each distribution. For both C iv and Si iv BAL troughs, the distributions of ∆EW have a stronger central peak, and both of the distributions are weaker than the Gaus- sian distributions for ∆EW 2–7 A.˚ Similarly, the ∆EW/ EW distributions for C iv | |≈ h i and Si iv BAL troughs are stronger for ∆EW/ EW < 0.2 and weaker for a range of | h i| ∆EW/ EW 0.2–0.6 compared to the Gaussian distributions. | h i| ≈ A non-Gaussian distribution of EW variations on timescales of more than 1 yr may appear as a result of the superposition of several Gaussian distributions, each of which characterizes the EW variations in a small time interval (see Figure 3.10). In order to test this hyphothesis, we ran a Lilliefors normality test for the ∆EW and ∆EW/ EW h i distributions in a small time interval. We selected the time interval of 2.0–2.5 yr, since a large number of BAL troughs are sampled in this range (see Figures 3.10 and 3.11); 156 C iv and 90 Si iv BAL troughs are sampled. The Lilliefors test showed that the ∆EW and ∆EW/ EW distributions for C iv and Si iv BAL-trough variations on ∆t = 2.0–2.5 yr h i min,1 are also non-Gaussian distributions (significance level of >95%). We have also investigated the ∆EW and ∆EW/ EW distributions for C iv and Si iv h i BAL troughs on timescales of ∆tmin < 1 yr. Performing the triples and Lilliefors tests, we found that the ∆EW and ∆EW/ EW distributions are also symmetric and non-Gaussian h i for both C iv and Si iv BAL troughs (significance level of >99.9%). 70 3.4.5 EW Variations as a Function of BAL-Profile Properties In this section, we investigate EW variations of BAL troughs as a function of BAL- profile properties, such as average EW, velocity width, depth, and centroid velocity. Al- though Gibson et al. (2008) found no significant correlation between the magnitude of EW variations and the average EW of C iv BAL troughs, we search for correlations for both C iv and Si iv BAL troughs with a much larger sample for variations in three different timescale ranges. Figure 3.15 shows ∆EW as a function of EW for short (< 1 yr), mod- h i erate (1–2.5 yr), and long (> 2.5 yr) timescales. Here the average EW, EW , indicates h i the average of the measured EWs of each BAL trough in two-epoch spectra. The number of data points used is given in the lower right of each panel. We illustrate with dotted curves where the EW variation is equal to the average EW. We search for correlations us- ing the Spearman rank-correlation test between ∆EW and EW . The test results show | | h i that the correlations are highly significant (>99%) for C iv BAL-trough EW variations on short, moderate and long timescales. The ∆EW for Si iv BAL troughs also likely | | correlates with EW on moderate timescales with a significance level of >95%. We do h i not detect significant correlations for variations of Si iv BAL troughs on short and long timescales, although these cases have the poorest trough statistics. These results suggest that weak BAL troughs tend to have small EW variations, an expected finding given that BAL troughs cannot weaken by more than their initial EWs measured in the first-epoch spectra. Remarkably, it is apparent in Figure 3.15 that weak BAL troughs also do not rise above the equality line. In addition, Figure 3.15 indicates that the EW variations of weak BAL troughs can be close to their average EWs. Studies by Lundgren et al. (2007) and Gibson et al. (2008) suggested a significant correlation between the fractional variation of EW, ∆EW/ EW , and EW . To investigate h i h i this correlation in a large sample, in Figure 3.16 we show ∆EW/ EW as a function h i of EW for C iv and Si iv BAL troughs for three timescales. In agreement with these h i previous studies, we find that the correlations between ∆EW/ EW and EW for C iv | h i| h i BAL troughs are highly significant (>99.9%) for all three timescale ranges. The test results demonstrate that the correlations between ∆EW/ EW and EW for Si iv BAL | h i| h i troughs are also highly significant (> 99.9%) for variations on moderate and long timescales. The significance level of the correlation is 98.4% for variations of Si iv troughs on short timescales. Figure 3.16 shows that the fractional variations of small-EW BAL troughs are larger compared to large-EW BAL troughs. Consistent with our results, the studies by Lundgren et al. (2007) and Gibson et al. (2008) found the ∆EW/ EW vs. EW | h i| h i correlation. However, they likely did not see significant correlation for ∆EW vs. EW | | h i due to the small sample size. Given that the width and depth of a BAL trough determine its EW, we assess the contributions of these two components to the significant correlation found between ∆EW | | and EW for C iv BAL troughs. We search for correlations between BAL-trough width, h i 71 ∆t < 1 yr ∆t = 1−2.5 yr ∆t > 2.5 yr min min min 10 0 −10 C IV ˚ A) ( 82 220 126 EW 10 ∆ log L BOL 0 Si IV −10 56 124 55 1 10 1 10 1 10 ˚ Fig. 3.15: EW variation, ∆EW, vs. average EW over the two relevant epochs, EW , h i for C iv (blue) and Si iv (red) BAL troughs for three different timescales as labeled. The x-axis is logarithmic. The black dotted curves denote where ∆EW is equal to EW . The | | h i number of data points are given in the lower right of each panel. 72 ∆t < 1 yr ∆t = 1−2.5 yr ∆t > 2.5 yr min min min 1 0 −1 C IV 82 220 126 1 10 EW / 0 −1 Si IV 56 124 55 1 10 1 10 1 10 ˚ Fig. 3.16: Fractional EW variation, ∆EW/ EW , vs. average EW over the two relevant h i epochs, EW , for C iv (blue) and Si iv (red) BAL troughs for three different timescales as h i labeled. The x-axis is logarithmic. The number of data points are given in the lower right of each panel. 73 ∆v, and BAL-trough depth, d (defined in Section 3.3.2), vs. ∆EW for variations on BAL | | short, moderate, and long timescales (see Figure 3.17). The Spearman test results show significant (>99.9%) correlations between ∆EW | | and ∆v for all three timescale ranges indicating that wider BAL troughs tend to vary more than narrower ones; this is perhaps as expected since wider BAL troughs might have a better chance of containing variable regions. Unlike BAL-trough width, we find no significant correlation between ∆EW and d . However, we do note from Figure 3.17 | | BAL that the EWs of the deepest BAL troughs (those with dBAL > 0.6) appear to vary less than shallower ones. An even larger sample will be required to investigate this behavior reliably since we have limited trough statistics at large values of dBAL. This apparent behavior could be due to more saturated absorption at large dBAL values; BAL absorption is often saturated and, due to partial covering, saturated BALs are usually found as non-black absorption (e.g., Hamann 1998; Arav et al. 1999b). The ∆EW distribution as a function of dBAL suggests that shallow BAL troughs on average increase their EWs and deep BAL troughs on average decrease their EW on moderate and long timescales (see Figure 3.17). We search for a correlation between ∆EW and dBAL using the Spearman test and find likely correlations for both moderate and long timescales (P = 2.6% and P = 5.6%, respectively). To show this trend, we also compare the mean ∆EW values for BAL troughs that are deeper and shallower than the median BAL trough depth. The median BAL-trough depth is 0.32 for moderate timescales and 0.31 for long timescales. We found the mean ∆EW values to be 0.95 0.30 A˚ for ± BAL troughs with d < 0.32 and 0.42 0.34 A˚ for BAL troughs with d > 0.32 on BAL − ± BAL moderate timescales. Similarly, for variations on long timescales the mean ∆EW values are 0.94 0.29 A˚ and 0.32 0.36 A˚ for BAL troughs shallower and deeper than d = 0.31, ± − ± BAL respectively. These results indicate that the statistical variation of a BAL trough depends not just on its EW and on the timescale between spectra, but also on its depth. It has been found in previous studies that weak BAL troughs generally can achieve higher velocities than strong BAL troughs (e.g., Gibson et al. 2009; Capellupo et al. 2011; Filiz Ak et al. 2012); our data also show this trend. Considering our results indicate that weak BAL troughs tend to have smaller EW variations, we search for correlations between ∆EW and central velocity of a BAL trough, v (see Section 3.3.2 for definition). | | cent Figure 3.17 displays ∆EW for C iv BAL troughs as a function of vcent for variations on three different timescales. Spearman test results show no significant correlations for variations on short, moderate, and long timescales. BAL variability has been noted to be fractionally stronger among weaker troughs (e.g., Lundgren et al. 2007; Gibson et al. 2008; Capellupo et al. 2011). However, since the average outflow velocity is higher for the population of weak BAL troughs than for strong troughs, this finding could, in principle, ultimately be a velocity effect. Owing to limited sample sizes, it has been difficult to assess if trough weakness or trough velocity is the primary driver of increased fractional BAL variability (e.g., Capellupo et al. 2011; Filiz Ak et al. 2012). To assess this with our larger sample, in Figure 3.18 we plot ∆EW/ EW vs. h i 74 15 ∆ t < 1 yr ∆ t = 1−2.5 yr ∆ t > 2.5 yr C IV min min min 10 ˚ A) ( 5 EW 0 ∆ −5 −10 −15 0 10 20 10 20 10 20 ∆v (103 km s−1) 15 10 ˚ A) 5 ( 0 EW ∆ −5 −10 −15 0.25 0.5 0.75 0.25 0.5 0.75 0.25 0.5 0.75 d BAL 15 10 ˚ A) ( 5 EW 0 ∆ −5 −10 −15 −30 −20 −10 −30 −20 −10 −30 −20 −10 v (103 km s−1) cent Fig. 3.17: C iv BAL-trough EW variation, ∆EW, as a function of BAL-trough width, ∆v (top panels), average depth of BAL troughs, dBAL (middle panels), and centroid velocity, vcent (bottom panels) for variations on three different timescales as labeled. Vertical blue lines in the middle panels indicate the median dBAL values and horizontal dashed red lines show the average ∆EW values for BAL troughs. 75 both EW and v . Spearman test results show that there is a clear correlation between h i cent ∆EW/ EW and EW (significance level of > 99.9%), but no significant correlation | h i| h i between ∆EW/ EW and v . Therefore, our results establish that trough weakness is | h i| cent indeed the primary driver of increased fractional BAL variability. We also search for correlations between ∆EW and ∆v, d , and v for Si iv | | BAL cent BAL troughs on three different timescales using Spearman tests. The test results show highly significant (>99.9%) correlations between ∆EW and ∆v for Si iv BAL troughs on | | all three timescales. Similar to the results for C iv BAL troughs, we found no significant correlation between ∆EW and d or ∆EW and v for Si iv BAL troughs. | | BAL | | cent 2 C IV 1 0 EW/ −2 0 20 40 60 80 −25 −20 −15 −10 −5 ( A)˚ v (103 km s−1) Fig. 3.18: C iv BAL-trough ∆EW/ EW as a function of EW (left) and v (right) for h i h i cent variations on timescales of more than 1 yr. 3.4.6 Comparison of C iv vs. Si iv EW Variations Owing to the differences in ionization potentials (45.1 eV for Si iv and 64.5 eV for C iv), abundances, and consequently optical depths, comparisons of BAL-trough variations between C iv and Si iv BAL troughs have an important role in assessing possible causes of BAL variations. To investigate variations of C iv and Si iv BAL troughs from the same absorbing material, we compare EWs and EW variations of BAL troughs of these two ions at corresponding velocities. As is well known from previous studies (e.g., Weymann et al. 1991; Barlow 1993; Gibson et al. 2009, 2010; Capellupo et al. 2011, 2012), C iv BAL troughs are not always accompanied by Si iv troughs. Therefore, we search for a C iv BAL trough in overlapping velocity ranges for each Si iv BAL trough in two-epoch spectra of each quasar that sample 76 iv minimum timescales of more than 1 yr, ∆tmin,1. Given that the Si BAL region can be contaminated by other lines between 13000 and 30000 km s−1 (see Section 3.4.1), in − − this comparison we search for accompanying C iv BAL troughs for a total of 136 Si iv BAL troughs with 3000 > v > 13000 km s−1. − cent − Figure 3.19 compares the average EW values, EW , for C iv and Si iv BAL troughs h i in overlapping velocity ranges. Consistent with previous studies (e.g., Gibson et al. 2009, 2010; Capellupo et al. 2012), we find that Si iv absorption tends to be weaker than C iv absorption present at the same velocities. The vast majority of Si iv BAL troughs ( 96%, ≈ 130 of 136) have EWs smaller than the corresponding C iv troughs. 50 ˚ A) ( 40 C IV 30 10 10 20 30 40 Fig. 3.19: Comparison of C iv and Si iv BAL-trough EWs for troughs having corresponding velocities. The dashed line indicates equal strengths of these two absorption troughs. C iv BAL troughs tend to be stronger than Si iv BAL troughs. A number of previous studies have searched for correlations between the variations of BAL troughs of different ions. Gibson et al. (2010) studied variations of nine C iv and Si iv BAL troughs. Although they were not able to find any correlation between EW variations of C iv and Si iv BAL troughs, they showed a strong correlation for fractional variations (see their Figures 11 and 12). Gibson et al. (2010) showed that fractional variations of Si iv absorption tend to be larger than those of C iv. Capellupo et al. (2012) monitored the C iv and Si iv BAL variability relationship using multi-epoch observations of a sample of 24 quasars. Although they were unable to demonstrate any significant correlation between 77 absolute variations of C iv and Si iv troughs (see their Figures 6 and 7), they noted a weak trend toward greater fractional change for Si iv. Figure 3.20 presents comparisons of ∆EW and fractional ∆EW measurements for C iv and Si iv BAL troughs in overlapping velocity ranges. Consistent with previous stud- ies, the EW variations of C iv and Si iv BAL troughs almost always occur in the same direction, and both the ∆EW and ∆EW/ EW of C iv and Si iv BAL troughs are clearly h i correlated. The Spearman test results demonstrate that the correlations are highly sig- nificant (> 99.9%) for both measurements. We also found that both the ∆EW and | | ∆EW/ EW of C iv and Si iv BAL troughs are strongly correlated (at > 99.9% signifi- | h i| cance), indicating that while the BAL troughs almost always vary in the same direction, the magnitudes of the variations are also correlated. 15 1 10 0.5 5 C IV ˚ A) ( C IV 0 0 EW ∆ −5 EW/ −15 −1 −15 −10 −5 0 5 10 15 −1.5 −1 −0.5 0 0.5 1 1.5 ∆EW ˚ ∆EW/ Fig. 3.20: Comparison of the variability of C iv vs. Si iv BAL troughs. The left panel compares EW variations for BAL troughs having corresponding velocities, and the right panel compares fractional EW variations. The light blue circles indicate Si iv and C iv BAL troughs with similar EWs as defined in the text. The dashed line in each panel has a slope of unity and zero offset. C iv and Si iv trough variations are clearly correlated; the solid black lines in each panel show the best-fit relations. As is apparent from Figure 3.20, both the ∆EW and ∆EW/ EW correlations have h i significant intrinsic scatter. To determine the relationship between EW variations of C iv and Si iv BAL troughs, we use a Bayesian linear-regression model that considers the in- trinsic scatter of the sample (Kelly 2007). The algorithm estimates regression parameters using random draws from the posterior distribution. We calculate the mean and standard deviation of model parameters that are found using 10000 random draws from our sam- ple. Using the linear-regression parameters from this algorithm, we found the following 78 relations: ∆EW = (1.223 0.106) ∆EW (0.038 0.271) (3.8) C IV ± × Si IV − ± ∆EW ∆EW = (0.551 0.032) (0.005 0.012) (3.9) EW ± × EW − ± h i C IV h i Si IV The standard deviation of the intrinsic scatter is 3.034 A˚ for the ∆EW and 0.123 for the ∆EW/ EW distributions between C iv and Si iv BAL troughs. h i Equation 3.8 shows that the slope of the relationship between ∆EW for C iv and Si iv BAL troughs is close to unity. The dispersion of the data points increases for large variations of C iv BAL troughs, indicating a possible non-linear component of correlation, especially for ∆EW & 10 A.˚ The ∆EW/ EW relation between C iv and Si iv BAL | C IV| h i troughs shows that the fractional EW variation of C iv BAL troughs is about half of the fractional EW variation of Si iv BAL troughs at a corresponding velocity. This result is consistent with the findings of Gibson et al. (2010) and Capellupo et al. (2012). In Section 3.4.5, we showed that weak BAL troughs tend to have larger fractional EW variability compared to strong BALs. In order to assess if the fractional EW of Si iv BAL troughs is more variable because, in general, Si iv BAL troughs are weaker than C iv BAL troughs (see Figure 3.19), we examine Si iv and C iv BAL troughs with similar EWs. We select a subset of BAL troughs where 0.8 < EWSi IV / EWC IV < 1.2 (a total of 17 cases) and repeat the linear-regression modeling. Analogous to Equation 3.8, we found the relation between ∆EW for these similar-strength C iv and Si iv BAL troughs to be ∆EW = (0.486 0.267) ∆EW + (0.159 0.413), indicating that Si iv troughs C IV ± × Si IV ± vary more in EW compared to C iv troughs of matched EW. Unfortunately, however, we could not derive a useful analog to Equation 3.9 since the fitted slope was poorly constrained by the data, and thus further work is needed to address this matter. Physically, we do expect the fractional variations of Si iv to be usually equal to or larger than those of C iv. BAL troughs have lower column densities and covering factors in Si iv than in C iv (e.g., Baskin et al. 2013). This situation arises primarily because the peak ionization fraction of Si iv is a factor of 2.5 lower than that of C iv and occurs at a slightly lower ionization parameter (e.g., Figure 4 of Hamann et al. 2000). Thus, C iv and Si iv absorption do not arise from exactly the same parts of an absorbing structure. When an absorbing cloud has a column high enough for the ionization inside it to decrease to the level where C iv is found, the cloud will produce C iv absorption; when the cloud has an even higher column, then both C iv and Si iv absorption will be produced. As a consequence, C iv and Si iv absorption are likely to be on parts of their curves of growth where we see either EW changes in C iv with no detectable Si iv absorption, small EW changes in saturated C iv and large EW changes in unsaturated Si iv (i.e., larger fractional change in Si iv), or small EW changes in both C iv and Si iv due to saturation (equal 79 fractional changes). The foregoing reasoning applies regardless of the origin of variations in C iv and Si iv absorption along the line-of-sight. Our data do rule out a scenario where all trough variations are due to the motion across the line-of-sight of clouds of uniform optical depth. 3.4.7 Coordination of EW Variations in BAL Quasars with Multiple Troughs Quasars with multiple BAL troughs of the same ion provide an opportunity to inves- tigate connections between distinct BAL troughs at different velocities. Using observations of BAL quasars with multiple troughs, Filiz Ak et al. (2012) demonstrated that variations in the EWs of multiple C iv BAL troughs are strongly correlated; when one BAL trough in a quasar spectrum disappears the other troughs present usually (11 out of 12 BAL troughs) weaken even for velocity offsets as large as 10000–15000 km s−1. Consistent with this, Capellupo et al. (2012) show that BAL variations at different velocities in the same ion almost always show coordinated variations. For the purpose of investigating coordinated variations of multiple BAL troughs of the same ion, we select 107 quasars with multiple C iv BAL troughs from our main sam- ple. We compare EW variations of the lowest-velocity BAL trough with those of the other BAL troughs (i.e., BAL troughs at higher velocities) present in the same pair of spectra. We detect a total of 137 higher-velocity BAL troughs in the spectra of these 107 quasars. Figure 3.21 shows the EW variation of the lowest-velocity BAL trough, ∆EWlow vel, as a function of EW variations for higher-velocity distinct BAL troughs, ∆EWhigh vel, for vari- ations on timescales of more than 1 yr. Similarly, we compare fractional EW variations for the lowest and higher-velocity BAL troughs in Figure 3.21. Variations of distinct troughs are correlated, although there is substantial scatter in the correlations. The Spearman test shows that both of the apparent correlations for C iv BAL troughs are highly (> 99.9%) significant. In order to assess correlations for magnitudes of ∆EW and ∆EW/ EW between the h i lowest-velocity and higher-velocity BAL troughs, we run the Spearman test again using the absolute values of the measurements and found highly significant (> 99.9%) correlations. To quantify the relationship between EW variations of the lowest-velocity and higher-velocity BAL troughs, we use the Bayesian linear regression model of Kelly (2007), considering the intrinsic scatter of the sample. Using this algorithm, we found the following relations: ∆EW = (0.674 0.094) ∆EW + (0.141 0.246) (3.10) high vel ± × low vel ± ∆EW ∆EW = (0.996 0.103) + (0.018 0.031) (3.11) EW ± × EW ± h i high vel h i low vel 80 10 ˚ A) ( 5 high vel 0 EW −1 ∆ Velocity offset (km s ) −5 0 − 5600 5600 − 8900 8900 − 12000 −10 >12000 −10 −5 0 5 10 ∆EW ˚ low vel(A) 1 high vel 0 EW/ −1.5 −1 −0.5 0 0.5 1 1.5 ∆EW/ Fig. 3.21: Comparison of EW variations (upper panel) and fractional EW variations (lower panel) of distinct troughs for BAL quasars with multiple C iv troughs for variations on timescales of more than 1 yr. Colors indicate the velocity separation between troughs; the relevant velocity ranges have been chosen to provide an equal number of data points in each range. The solid black lines in each panel show the best-fit relations. Variations of distinct troughs are correlated for C iv, although there is substantial scatter in the correlations. 81 The standard deviation of the intrinsic scatter is 2.769 A˚ for the ∆EW and 0.324 for the ∆EW/ EW distributions for multiple C iv BAL troughs. h i Equation 3.10 indicates that BAL troughs at higher velocities tend to have smaller EW variations compared to the lowest-velocity BAL troughs. This result is expected considering that small-EW BAL troughs can achieve higher velocities, and they also tend to have small ∆EW values (see Section 3.4.5 and Figure 3.15). The slope of Equation 3.11 is close to unity, indicating that multiple BAL troughs present in the same two-epoch spectra show similar fractional EW variations. This result is even more interesting given that weak BAL troughs tend to have larger ∆EW/ EW compared to strong ones (see h i Section 3.4.5 and Figure 3.16). Owing to their large fractional variations, disappearing or emerging BAL troughs (see Section 3.5.3) stand out in the lower panel of Figure 3.21. The higher-velocity BAL trough of the quasar SDSS J143948.06+042112.8, an emergence candidate with ∆EW/ EW = 1.46, is found as an outlier at the top-left of the plot. The other emer- h ihigh vel gence candidate, the higher-velocity BAL trough of the quasar SDSS J151312.41+451033.9, is found at the extreme of the correlation with ∆EW/ EW = 1.34. Similarly, h ihigh vel the higher-velocity BAL trough of SDSS J095901.24+550408.2 that satisfies the disap- pearance criteria used in Filiz Ak et al. (2012) is shown at the lower-left of the plot (∆EW/ EW = 1.5). h ihigh vel − Table 3.6 presents the number of additional BAL troughs that vary in the same or opposite direction as the lowest-velocity BAL trough. We found that 78.1 8.3% of ± the additional C iv BAL troughs show correlated variations with the lowest-velocity BAL trough. A combinatorial probability calculation shows that the probability of obtaining such a result for independently varying BAL troughs is 10−12. ≈ To determine the velocity range over which coordinated variations of multiple troughs occur, we calculate the velocity offset of those BAL troughs at higher velocities from the lowest-velocity BAL trough present in the same pair of spectra. Figure 3.22 shows the per- centage of additional C iv BAL troughs at a given velocity offset with increasing strength (∆ EW > 0) and decreasing strength (∆ EW < 0) in comparison with the ∆EW varia- tion of the lowest-velocity BAL trough. We find that even BAL troughs separated from the lowest-velocity trough by 15000–20000 km s−1 show generally correlated variations. Figure 3.21 indicates that BAL troughs with small velocity offsets plausibly have similar scatter about the correlations as troughs with large velocity offsets. To assess if troughs with small velocity offsets show better coordination than those with large velocity offsets, we investigate the fraction of BAL troughs showing coordinated variations as a function of velocity offset using the velocity ranges given in Figure 3.21. We found that the fraction of higher-velocity BAL troughs showing coordinated variations is constant at 78% within ≈ the error bars, indicating no strong dependence upon velocity offset. We also examine the coordination of multiple troughs for the Si iv transition. Similar to C iv BAL troughs, the ∆EW and ∆EW/ EW variations for Si iv are correlated; the h i Spearman test indicates that the significance level is > 99.9% for both correlations. As for 82 100 50 1778534928410121021001 0 50 ∆ 100 Lowest velocity trough EW > 0 100 Lowest velocity trough ∆EW < 0 50 0 1967467743342232 Percentage of additional BAL Troughs 50 100 5000 10000 15000 20000 Velocity offset from the lowest velocity trough (km s−1) Fig. 3.22: Percentage of additional C iv BAL troughs at a given velocity offset with in- creasing strength (above zero level in each panel; ∆ EW > 0) and decreasing strength (below zero level; ∆ EW < 0). The upper panel is for cases where the lowest-velocity BAL trough strengthens (∆EWlow vel > 0), and the lower panel is for cases where it weakens (∆EWlow vel < 0). The numbers for each bar show the total number of additional BAL troughs found at the relevant velocity offset. Even BAL troughs separated from the lowest velocity trough by 15000–20000 km s−1 show generally correlated variations. 83 C iv BAL troughs, multiple Si iv BAL troughs also tend to vary in the same direction. The Si iv BAL-trough region, however, is subject to contamination by some other lines in the velocity range of 13000–30000 km s−1 (see Section 3.4.1). Considering such contamination, we do not perform further examinations for multiple Si iv troughs. 3.4.8 EW Variations as a Function of Quasar Properties In this section, we study BAL-trough EW variations as a function of quasar prop- erties, including luminosity, Eddington luminosity ratio, redshift, and radio loudness. 3.4.8.1 Luminosity Previous studies (e.g., Kaspi et al. 2005; Bentz et al. 2009) have investigated the relationship between luminosity and broad-line region (BLR) size in AGN. In these studies the derived power-law relation indicates that the size of the BLR increases with increasing quasar luminosity. Given that the BAL region often lies outside the BLR (e.g., BALs often absorb BLR emission), this relation may also be indicative of an increasing size and decreasing orbital (transverse) velocity of the BAL region with increasing luminosity. A dependence of the transverse BAL velocity, and/or continuum source region size, on luminosity can potentially change the variability of BAL outflows as a function of luminos- ity. Furthermore, Laor & Brandt (2002) found a correlation between outflow velocity and quasar luminosity, indicating that BAL troughs in luminous quasar spectra can achieve higher velocities. Later, a study by Ganguly et al. (2007) investigated outflows and the physical properties of quasars using a sample of 5088 quasars. In addition to finding consis- tent results with Laor & Brandt (2002) on the luminosity-outflow relation, Ganguly et al. (2007) showed that the fraction of BAL quasars (relative to all quasars) increases with increasing luminosity. Therefore, we investigate BAL-trough EW variations as a function of the quasar bolometric luminosity, LBol. As mentioned in Section 3.2.2 LBol values are taken from Shen et al. (2011). In Figure 3.23, we show the dependence of ∆EW for C iv and Si iv BAL troughs upon LBol for variations on three different timescales. In each panel of Figure 3.23, we show standard-deviation curves calculated using a sliding window for luminosity-ordered data points where the mean EW error in each window is statistically removed from the curve. The number of data points contained in each window is given in the lower left of each panel of Figure 3.23. If BAL variability decreased with luminosity, one would expect to see a larger scatter of the data points at low luminosities as a result of EW variations in both directions (and vice versa for increasing variability with luminosity). In Figure 3.23, the standard deviation curves for C iv and Si iv on moderate and long timescales may be indicative of BAL variability decreasing with luminosity. To assess formally any correlation between L and ∆EW , we use the Spearman Bol | | test. Table 3.7 lists the test probability results along with the number of data points 84 15 ∆t < 1 yr ∆t = 1−2.5 yr ∆t > 2.5 yr C IV min min min 10 5 0 −5 −10 −15 8 22 12 ˚ A) ( EW Si IV ∆ 5 0 −5 6 12 6 46.5 47 47.5 46.5 47 47.5 46.5 47 47.5 log L (erg s−1) Bol Fig. 3.23: EW variability of C iv (upper panels) and Si iv (lower panels) BAL troughs as a function of quasar bolometric luminosity, LBol, for three different rest-frame timescales as labeled. The black curves in each panel show the running rms computed with a sliding window; the window width, in terms of number of troughs, is given in the lower left of each panel. If BAL variability decreased (increased) strongly with luminosity, one would expect to see a larger (smaller) scatter of the data points at low luminosities. We do not find significant evidence for such behavior on any of the sampled timescales, at least over the order-of-magnitude in luminosity with significant coverage. 85 in each timescale range. Generally, we do not find significant evidence for correlations between L and ∆EW or L and ∆EW/ EW . The one exception is for C iv troughs Bol | | Bol | h i| on moderate timescales (99.8% significance); this is also the case where we have the best trough statistics. However, given the number of trials in our correlation tests, we do not regard this one case as strong evidence for luminosity dependence. 3.4.8.2 Eddington Luminosity Ratio and SMBH Mass Proga & Kallman (2004) suggested that the presence of radiatively driven disk winds is highly sensitive to the ratio of the quasar bolometric luminosity to the Eddington luminosity, LBol/LEdd. Therefore, quasar BAL properties are plausibly expected to depend upon LBol/LEdd. Indeed, studies by Laor & Brandt (2002) and Ganguly et al. (2007) found a correlation between the outflow velocity of quasar winds and LBol/LEdd. To assess the effect of LBol/LEdd on the variability of the BAL region, we investigate correlations between BAL-trough EW variations and LBol/LEdd. Shen et al. (2011) calculated LBol/LEdd values using the virial black-hole mass which is estimated using the C iv emission line for z 1.9 ≥ quasars. As is well known (e.g., Shen et al. 2008, 2011; Shen 2013, and references therein), LBol/LEdd calculations using the C iv emission line show a larger scatter compared to other emission lines such as Hβ and Mg ii. Therefore, uncertainties in the LBol/LEdd measurement may hide any underlying correlations. Figure 3.24 shows ∆EW for C iv and Si iv BAL troughs as a function of LBol/LEdd on short, moderate and long timescales. Similar to Figure 3.23, we show the standard- deviation curves computed with a sliding window in each panel of Figure 3.24. As for lumi- nosity, if BAL variability decreased (increased) with Eddington luminosity ratio, one would expect to see a larger (smaller) scatter of the data points at low Eddington-normalized lu- minosities. However, the standard-deviation curves do not indicate such a trend for C iv and Si iv BAL troughs on the sampled timescales. We run Spearman tests to assess connections between BAL EW variations and LBol/LEdd. The test results generally show no evidence of significant correlations between L /L and ∆EW or L /L and ∆EW/ EW for C iv and Si iv BAL troughs Bol Edd | | Bol Edd | h i| on the sampled timescales (see Table 3.7). However, as for luminosity, the small probabil- ity for C iv troughs on moderate timescales might be indicative of a correlation between L /L and ∆EW . This result is consistent with the test result between L and Bol Edd | | Bol ∆EW considering the strong relation between L and L /L . | | Bol Bol Edd In addition, we assess any correlation between EW variations of BAL troughs and SMBH mass estimates from Shen et al. (2011). The Spearman test results show no evidence of significant correlations between M and ∆EW or M and ∆EW/ EW for C iv BH | | BH | h i| and Si iv BAL troughs on the sampled timescales (see Table 3.7). 86 ∆t < 1 yr ∆t = 1−2.5 yr ∆t > 2.5 yr 15 min min min C IV 10 5 0 −5 −10 −15 8 21 12 ˚ A) ( −1 0 −1 0 log L/L edd EW Si IV ∆ 5 0 −5 6 12 5 −1 0 −1 0 −1 0 log L / L Bol Edd Fig. 3.24: Same as Figure 13 but for Eddington-normalized luminosity, LBol/LEdd. We do not find any significant evidence for changes in BAL variability with LBol/LEdd. 87 3.4.8.3 Redshift It is also of interest to assess any dependence of BAL variability upon redshift; e.g., since Allen et al. (2011) reported a relation between the fraction of BAL quasars and redshift. Therefore, we investigate correlations between quasar redshifts and EW variations on short, moderate, and long timescales. Plots similar to Figures 3.23 and 3.24 generally do not show significant relations with redshift, consistent with the results in Table 3.7. 3.4.8.4 Radio Loudness Do BAL troughs in radio-quiet and radio-loud quasars vary differently? Given that the presence of powerful radio jets in a quasar appears to affect its wind properties (e.g., Becker et al. 2000; Shankar et al. 2008; Miller et al. 2009, 2012; Welling et al. 2013), different BAL variability might be expected for radio-loud quasars. To address this issue, we first examine ∆EW and ∆EW/ EW as a function of R h i for BAL variations on timescales of more than 1 yr; here R is the radio-loudness parameter calculated by Shen et al. (2011) using VLA FIRST observations (see Section 3.2.2). Radio emission is detected from 33 of our main-sample quasars, nine of which have R> 100 and thus are radio-loud. We identified 40 C iv and 28 Si iv BAL troughs in multi-epoch spectra of these 33 quasars. In Figure 3.25, we show the distribution of ∆EW for the 40 C iv BAL troughs as a function of R. A Spearman test shows no indication of a significant correlation between ∆EW and R. Similarly, we find no significant correlation between ∆EW/ EW | | | h i| and R. We repeat the correlation tests for those nine radio-loud quasars with R> 100 and found no significant correlations. Second, we compare BAL-trough EW variations on matched timescales for radio- detected (R & 1) and radio-non-detected (R . 1) quasars. We randomly select a sample of C iv BAL troughs of radio-non-detected quasars that have matching timescales with the 40 C iv BAL troughs, and then run a two-sample KS test comparing the ∆EW distributions of these two samples. The KS test results show no inconsistency between the two ∆EW distributions. Similarly, we do not find significant inconsistency from the comparison of the ∆EW/ EW distributions. h i We also repeat the matched-sample selection and KS tests for those nine radio-loud quasars with R> 100. We find no inconsistency between BAL-trough variability of radio- loud and radio-non-detected quasars from the comparison of the ∆EW and ∆EW/ EW h i distributions for C iv BAL troughs. 3.5 Discussion In Section 3.4, we presented our main observational results with generally limited physical discussion. In this section, we therefore examine the physical implications of our most notable findings for quasar winds. 88 Number of BAL Troughs 50 100 15 10 5 ˚ A) ( EW 0 ∆ −5 −10 −15 1.0 10 100 R Fig. 3.25: EW variation for C iv BAL troughs as a function of radio-loudness parameter, R, for variations on timescales of more than 1 yr. Individual data points are plotted for quasars with radio detections in the FIRST catalog, while a histogram is shown along the left-hand side for quasars lacking radio detections. We do not find strong evidence for changes in BAL variability with R. 89 3.5.1 The Frequency of BAL Variations Our large sample demonstrates that C iv and Si iv BAL variations are common on multi-year rest-frame timescales. About 50–60% of both C iv and Si iv BALs vary detectably over 1–3.7 yr, with the exact derived fraction depending upon the approach taken in variable BAL-trough identification. The fraction of quasars showing BAL-trough variability is even higher, since some BAL quasars possess more than one trough. Using +4.9 +6.3 the two-epoch observations of our sample, we find that 62.2−4.6% (181/291) and 59.1−5.7% (107/181) of BAL quasars show C iv and Si iv trough variability, respectively. Our results are derived from a much larger sample of quasars than past work, and they also do not suffer from systematic uncertainty owing to multi-counting biases (see Footnote 2) or preferential observation of BAL quasars known to show trough variability (e.g., see Section 3.3 of Capellupo et al. 2013). However, our results are generally consistent with past studies. For example, Capellupo et al. (2013) found that 55% of their sample of BAL quasars ≈ showed C iv trough variability on a timescale of 2.5 yr (see their Figure 12). Gibson et al. (2008) found that 12 of their 13 ( 92%) BAL quasars showed C iv trough variability, ≈ although they sampled longer timescales of 3–6 yr where a higher percentage of variable BAL quasars is expected. The high incidence of BAL variability that we find on multi-year timescales is gen- erally supportive of models where most BAL absorption arises within about an order- of-magnitude of the radial distance of the launching region; i.e., 10–1000 light days (see Section 3.1). However, our results are still consistent with a significant fraction of BAL ab- sorption arising on larger scales. Indeed, some of the troughs in our sample remain remark- ably stable over multiple years in high-quality spectra (e.g., see Figure 3.26). This result is consistent with independent findings that some BAL absorption arises on 100–3000 pc ≈ scales (e.g., Faucher-Gigu`ere et al. 2012; Arav et al. 2013; Borguet et al. 2013), although lack of variability does not strictly require that the absorption arises on these size scales. In some cases, we find both remarkably stable as well as variable troughs in the same object (e.g., SDSS J141513.99+365412.2 and SDSS J225608.48+01557.7 in Figure 3.26), suggesting that BAL absorption may arise over a wide range of size scales. Those BAL troughs believed to be formed on 100–3000 pc scales should be intensively monitored for ≈ variability as an independent check of their distance from the SMBH. 3.5.2 Constraints Upon BAL Lifetimes In Section 3.4.3, we examined the behavior of ∆EW and ∆EW/ EW for BAL h i troughs as a function of rest-frame timescale. One implication of these results is that we can make basic order-of-magnitude estimates of the average “lifetime” of a BAL trough; here the lifetime represents the time over which a BAL trough is seen along our line-of-sight and not necessarily the physical lifetime of the absorbing structure (e.g., a structure could remain intact but move out of our line-of-sight). We use Equation 3.6 considering that a 90 SDSS J021646.94−092107.2 2.0 SDSS J073535.44+374450.4 1.5 z = 3.7 z = 2.8 λ ∆t = 774 days 1.5 ∆t = 881 days 1.0 1.0 Density F 0.5 Normalized Flux 0.5 CIV CIV 2.0 2.0 σ 0.0 0.0 N −2.0 −2.0 2.0 SDSS J141513.99+365412.2 2.0 SDSS J225608.48+010557.7 λ z = 2.9 z = 2.3 1.5 ∆t = 662 days 1.5 ∆t = 1008 days 1.0 1.0 Density F Normalized Flux 0.5 0.5 SDSS BOSS 5.0 2.0 σ 0.0 0.0 N −5.0 −2.0 1420 1440 1460 1480 1500 1520 1540 1560 1420 1440 1460 1480 1500 1520 1540 1560 ˚ ˚ Rest Wavelength ( A) Rest Wavelength ( A) Fig. 3.26: Two-epoch spectra from SDSS (red) and BOSS (black) of quasars with BAL troughs that remain remarkably stable over multiple years in the rest frame; in some cases additional distinct BAL troughs that vary are present as well. These are examples of quasars that may have some BAL absorption arising on large physical scales. The dashed vertical lines show the C iv emission-line rest wavelength. The lower section of each panel shows Nσ values for SDSS vs. BOSS observations where the dashed red lines show the 1σ level. ± 91 BAL trough will disappear when the magnitude of its EW variation is equal to its EW, ∆EW = EW. In our current sample the measured EWs are in the range 0.3–87.5 A˚ with | | a median of 10.9 A.˚ Given that the strengthening and weakening of BAL troughs occur at similar rates (see Section 3.4.4), solution of Equation 3.6 for ∆EW = 10.9 A˚ gives | | half of an average BAL lifetime (the other half corresponds to the BAL strengthening from 0 A˚ to 10.9 A).˚ Therefore, this approach reveals that t = 1600+2100 yr. The large h BALi −900 error bars of our lifetime estimate largely arise because of the significant difference between the sampled timescales and the derived lifetime. Although the measurements underlying Equation 3.6 span a factor of 103 in timescale, the lifetime appears a factor of 103 ≈ ∼ longer still (and thus, of course, the extrapolation used to estimate a lifetime needs to be treated with caution). Our estimate of the average BAL lifetime is consistent with the lower limits for BAL lifetime in Gibson et al. (2008, 2010) and Hall et al. (2011). We note that this average lifetime is long compared to the orbital time of the accretion disk at the expected wind launching radius of 10–100 light days (for a 109 M SMBH, t is 50 yr). ⊙ orb ∼ It is also long, or at best comparable to, the orbital time at the radius of the BLR of 1 ≈ light year (e.g., Kaspi et al. 2007; t 500 yr). orb ∼ One should keep in mind that BAL lifetimes derived with the above approach depend significantly upon EW, and thus the lifetimes derived here are in agreement with those estimated by Filiz Ak et al. (2012) using a qualitatively different approach. Using a sample of 21 examples of disappearing BAL troughs, Filiz Ak et al. (2012) found the average BAL lifetime to be about a century; the measured EWs of these 21 BAL troughs in the first- epoch spectra were 2.2–10.6 A˚ , with a median of 4.7 A.˚ Applying the approach based upon Equation 3.6, we find t = 60+38 yr for these 21 disappearing BAL troughs, consistent h BALi −22 with Filiz Ak et al. (2012). 3.5.3 Relation of BAL Disappearance and Emergence to General BAL Vari- ability In this section, we assess where BAL disappearance and emergence events lie within the distribution of BAL EW variability. Since our sample-selection criteria (see Sec- ′ tion 3.2.2) limit the redshift, SN1700, and BI more strictly than in Filiz Ak et al. (2012), our main sample includes only three of the disappearing C iv BAL troughs described in Filiz Ak et al. (2012). None of these three has a Si iv BAL trough at corresponding velocities at any epoch. Figure 3.14 displays the three examples of disappearing C iv BAL troughs. Filiz Ak et al. (2012) found that disappearing BAL troughs tend to have relatively small EWs in their first-epoch spectra; the three examples of disappearing C iv BAL troughs lie between 0 and 5 A˚ in the ∆EW distribution. In the distribution of fractional EW − variations, all three of the disappearing C iv BAL troughs lie at the negative extreme of the distribution, although they are not a distinct population. These examples of BAL dis- appearance indicate that disappearance is an extreme example of general BAL variability, rather than a qualitatively distinct phenomenon. 92 We have visually inspected the BAL troughs whose EWs decreased by at least a factor of five between the two epochs (i.e., BAL troughs with ∆EW/ EW < 1.33). h i − Our main sample includes one additional case of C iv BAL disappearance that satis- fies the disappearance criteria used in Filiz Ak et al. (2012); this trough, in the quasar SDSS J095901.24+550408.2, is plotted in the upper-right panel of Figure 3.14 at ∆EW/ EW = 1.5. Two of the main-sample quasars (SDSS J092522.72+370544.1 and h i − SDSS J112055.79+431412.5), whose Si iv BAL trough EWs decreased by at least a factor of five between the two epochs, can be classified as candidates for quasars with disappear- ing Si iv BAL troughs. The corresponding C iv BAL troughs of these two quasars do not disappear. Visual inspection of the BAL troughs whose EWs increased by at least a factor of five between the two epochs (i.e., BAL troughs with ∆EW/ EW > 1.33) shows that two of the h i C iv BAL troughs (for the quasars SDSS J143948.06+042112.8 and J151312.41+451033.9) and two of the Si iv BAL troughs (for the quasars SDSS J145045.42 004400.2 and − J160202.40+401301.4) can be classified as candidates for BAL-trough emergence. These emergence events lie at the positive extreme of the distribution of fractional EW variations, but again do not appear to be a distinct population. The detailed analysis of emergence is beyond the scope of this study, but we plan to address this topic in future work. 3.5.4 A Random-Walk Model for The Evolution of BAL Troughs Modelers of quasar winds have not yet been able to make quantitative predictions of how BAL-trough EWs should evolve over time, and thus it is not possible to use our observational data on this topic to test wind models directly. Therefore, as an alternative, we test and constrain a “toy” model where long-term EW variations occur as a series of discrete events. Assuming that a BAL-trough EW varies by a fixed amount δEW after a fixed time step δt, we use a simple one-dimensional (unbiased) random-walk model to characterize the EW evolution. Therefore, we assume that over a period of time, T , a BAL trough undergoes n = T/δt changes to its EW. For the purpose of defining T , we select our data to examine BAL-trough EW variations on timescales of 2.0–2.5 yr (see Section 3.4.4 for the motivation for this interval choice). Therefore, we assume each BAL trough evolved with the same number of steps over a time T = 2.25 yr (i.e., the mean of the sampled timescales in this interval). The ∆EW distribution for C iv BAL-trough variations on 2.0–2.5 yr timescales is shown in Figure 3.27. A random-walk model produces a binomial distribution for ∆EW determined by two parameters: the number of steps, n, and the size of each step, s. To determine the random-walk model that best matches our data, we produced a set of binomial distributions using different numbers of steps and step sizes, and performed a χ2 test between our data and each distribution. A random-walk model with n = 6 and s = 1.65 A˚ yields the minimum χ2 for C iv BAL troughs on timescales of 2.0–2.5 yr. The best-fitting parameters suggest that C iv BAL troughs in this model tend to evolve in a small number of steps 93 70 C IV ∆EW 60 Binomial Distribution Gaussian Distribution 50 40 30 Number of BAL Troughs 20 10 0 −14 −12 −10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 ˚ ∆EW (A) Fig. 3.27: Distribution of C iv BAL-trough EW variations, ∆EW, for variations on timescales of 2.0–2.5 yr and a binomial distribution calculated for a random-walk model with six steps and a step size of 1.65 A˚ . We bin our ∆EW data at a bin width of 2 A,˚ and calculate the counting errors following Gehrels (1986). The black dashed line shows the best-matching Gaussian distribution. 94 on 2.0–2.5 yr timescales. Figure 3.27 shows the binomial distribution that best matches the ∆EW distribution. The P (χ2,ν) = 64.3% for the best-fitting model indicates that the data are consistent with the random-walk model. We calculate confidence intervals for the model parameters using numerical ∆χ2 confidence-region estimation for two parameters of interest (see Section 15.6.5 of Press et al. 1992). In Figure 3.28, we show confidence intervals of 68.3%, 90%, and 99% for two parameters of interest. The best-fitting fixed time +46 step is δt = 137−53 days; we calculate uncertainties on this parameter at 68.3% confidence. A binomial distribution produced by a random-walk model closely approximates a Gaussian distribution for a large number of steps (n & 20); for a smaller number of steps the binomial distribution remains non-Gaussian. Consistent with the fact that the ∆EW distribution for C iv BAL troughs on 2.0–2.5 yr timescales is non-Gaussian (see Section 3.4.4), the results in Figure 3.28 show that our data on the ∆EW distribution for C iv BAL troughs reject any Gaussian approximation (i.e., large values of n are disfavored). As for the ∆EW distributions in Section 3.4.4, the binomial model distribution also has a stronger central peak compared to the Gaussian approximation. Similarly, we use a random-walk model to describe the evolution of Si iv BAL-trough EWs. The best-fitting model parameters for Si iv BAL-trough EW variations on timescales of 2.0–2.5 yr are n = 6 and s = 1.56 A˚ with P (χ2,ν) = 25.2%. The best random-walk model parameters for both C iv and Si iv BAL troughs are almost identical; this result is broadly consistent with the correlated C iv and Si iv BAL-trough variations discussed in Section 3.4.6. Using the best-fitting random-walk model (n = 6, s = 1.65 A)˚ calculated for C iv BAL-trough EW variations on timescales of 2.0–2.5 yr, we simulated EW variations for a sample of C iv troughs; we selected the simulated sample to have the same size as our main sample (428). Figure 3.29 shows a comparison between the simulations and observed EW variations on timescales of 5.9 hr to 3.7 yr. Given that the fixed time step is larger than the shorter timescales sampled by our data, we randomly select the start time of each simulated trough and preserved δt = 137 days for each step. As can be seen in Figure 3.29, the best-fitting random-walk model on timescales of 2.0–2.5 yr appears to represent acceptably the range of EW variations on a larger range of timescales. We also compared sliding-window standard deviation curves computed for the data and the model. These curves are consistent on timescales of 100–1400 days, and on shorter timescales the data are too limited for proper comparison. The simple random-walk model examined above should only be considered to be an approximation of a more complex reality; e.g., BAL EWs surely do not evolve via discrete steps after remaining constant for a fixed interval. Nevertheless, the basic apparent success of this model suggests that some of its features are of importance. For example, the derived characteristic timescale of 137 days may represent the length of time that BAL EWs ≈ tend to increase/decrease monotonically. This timescale also suggests that the distribution of BAL EW values will only approximate a Gaussian for monitoring periods of & 7.5 yr in the rest frame. 95 16 14 12 10 n 8 6 4 2 1.2 1.4 1.6 1.8 2 2.2 s Fig. 3.28: 68.3% (dark blue), 90% (light blue), and 99% (dark red) confidence intervals (corresponding to ∆χ2 = 2.30, 4.61, and 9.21) for the two parameters of the random-walk model describing C iv BAL-trough EW variations, ∆EW, on timescales of 2.0–2.5 yr. The black diamond indicates the best-fitting model with the number of steps, n = 6, and the size of each step, s = 1.65 A.˚ 96 15 C IV ∆EW Random−Walk Model 10 1.5 5 EW 0 ∆ −5 −10 −15 1 10 100 1000 ∆t (days) min Fig. 3.29: Comparison between EW variations for 428 C iv BAL troughs in our main sample and the best-fitting random-walk model. Vertical red bars show a timescale range of 2.0–2.5 yr. 3.5.5 Shielding Gas Changes as a Driver of Coordinated Multi-Trough Vari- ability? Our large sample has allowed us to establish, beyond question, the coordinated variability of multiple troughs in quasar spectra (Capellupo et al. 2012; Filiz Ak et al. 2012). This phenomenon is found for both C iv and Si iv troughs; e.g., for C iv the derived probability of this phenomenon being due to chance is 10−12. Our investigations of ≈ coordinated multi-trough variability show that it is present even for velocity offsets as large as 15000–20000 km s−1. Coordinated multi-trough variability might appear somewhat surprising in the con- text of quasar-wind models (see Section 3.1), and it can provide insights into the drivers of BAL variability more generally. Absorption troughs at widely separated velocities should be formed in outflow streamlines that are widely physically separated and thus largely unrelated. Some mechanism capable of acting over a wide range of radii must be present to enforce the coordinated multi-trough variability. Consideration of current quasar-wind models suggests a possible enforcement mechanism is substantial changes in the amount of “shielding gas” along the line-of-sight (e.g., Misawa et al. 2007; Filiz Ak et al. 2012); the shielding gas prevents BAL outflows from being overionized by nuclear extreme UV (EUV) and X-ray photons (e.g., Murray et al. 1995; Proga et al. 2000). Shielding-gas variability is likely according to quasar-wind simulations (e.g., Proga et al. 2000; Sim et al. 2010, 2012), and it may have been detected in a couple of BAL quasars (e.g., Gallagher et al. 2004; Saez et al. 2012). It can arise due to internal gas motions and/or accretion-disk rotation at the 97 relatively small radii where the shielding gas is located. Changes in the column density of shielding gas can increase or decrease the level of ionizing EUV/X-ray radiation reaching the BAL wind. In response to this variation, absorption components at different velocities can rise or fall in ionization level together, leading to coordinated multi-trough variations.3 Of course, this same mechanism might be applicable to single-trough BAL quasars as well and thus much of BAL variability in general. One method for testing this scenario would be to examine the continuum variability of quasars showing coordinated multi-trough variations. Ideally one would measure the relevant EUV/X-ray ionizing continuum directly, but measurements of the more straight- forwardly accessible rest-frame UV continuum might suffice. Unfortunately, limitations in the spectrophotometric calibration of the BOSS spectra do not allow this test to be implemented straightforwardly at present, but it should be possible in the future as BOSS calibration efforts proceed (Margala & Kirkby 2011; Suzuki 2013). It is also perhaps possible that EUV/X-ray luminosity variations could primarily drive the BAL variability rather than shielding-gas variations. This would require that some EUV/X-ray radiation is not blocked from reaching the wind by the shielding gas. 3.6 Summary and Future Work We have studied the variability of C iv and Si iv BAL troughs using a systematically observed sample of 291 BAL quasars. We have utilized 699 high-quality spectra of these quasars obtained by SDSS-I/II and BOSS. The main basic observational results of our study are the following: 1. Using a BAL-trough definition designed for the investigation of trough variability in multi-epoch observations, we have identified 428 distinct C iv and 235 distinct Si iv BAL troughs in the multi-epoch spectra of our 291 “main-sample” quasars. The sampled rest-frame timescales range between 5.9 hr and 3.7 yr with a median of 2.1 yr. See Section 3.3.2. 2. We identify variable BAL troughs using two different approaches and find, consistent with earlier work, that BAL variability is common on multi-year timescales. About 50–60% of both C iv and Si iv BAL troughs detectably vary on rest-frame timescales of 1–3.7 yr. Our large sample size also allows us to quantify the fraction of BALs that vary by a given amount in EW. The cumulative fraction of BAL troughs varying by a given ∆EW threshold drops with increasing ∆EW , but it remains significant | | | | even for ∆EW values as large as 5 A.˚ See Sections 3.3.3 and 3.4.1. | | 3The absorption depth in less-saturated regions of absorption would naturally be more responsive to ionization changes than that in highly saturated regions, and such velocity dependent saturation effects might explain how C iv and Si iv variations usually occur in relatively narrow portions of BAL troughs (see Section 3.4.2). 98 3. C iv BAL variability is found across a wide range of outflow velocity. The percentage of C iv BAL regions showing variability remains relatively constant at 30–40% at velocities from 3000 km s−1 to 25000 km s−1, and it then rises at higher velocities. − − The dependence of Si iv BAL variability upon outflow velocity is more difficult to constrain owing to contamination by unrelated emission lines. See Section 3.4.1. 4. We show, consistent with earlier work, that C iv and Si iv BAL variations usually occur in relatively narrow portions of BAL troughs (i.e., troughs do not vary mono- lithically in depth). The variable portions typically span < 30% of the trough velocity width. Narrow BAL troughs tend to have variable portions that span a larger fraction of their velocity width than broad BAL troughs. See Section 3.4.2. 5. The incidence of variability within a BAL trough does not depend significantly upon the relative velocity within the trough. See Section 3.4.2. 6. The EW and fractional EW variations of C iv and Si iv BAL troughs increase with sampled rest-frame timescale, and our large sample allows improved measurement of this effect over a range of 103 in timescale. We find that the rate-of-change of BAL ≈ EW is larger on 0.01–1 yr than on 1–3.7 yr timescales. See Section 3.4.3. 7. Even in this large sample, the distributions of EW variations and fractional EW vari- ations for C iv and Si iv BAL troughs show no detectable deviation from symmetry about zero. This indicates that BALs do not strengthen and weaken at significantly different rates. Both the ∆EW and ∆EW/ EW distributions appear non-Gaussian. h i See Section 3.4.4. 8. We have assessed, in several timescale ranges, correlations between BAL EW vari- ations and BAL-profile properties, including average EW, velocity width, depth, and outflow velocity. Weak BAL troughs tend to have smaller absolute EW vari- ations but larger fractional EW variations than strong troughs. Decomposing EW into velocity-width-based and depth-based components, BAL EW variability depends more strongly upon velocity width. We found no correlation between absolute EW variations and trough outflow velocity. See Section 3.4.5. 9. The ∆EW and ∆EW/ EW values of C iv BAL troughs correlate with those of h i Si iv troughs corresponding in velocity. The ∆EW correlation has significant scatter, perhaps suggesting a non-linear correlation. The ∆EW/ EW correlation shows that h i the fractional EW variation of C iv BAL troughs is about half that of Si iv troughs. See Section 3.4.6. 10. Our large sample clearly establishes that when a BAL quasar shows multiple troughs of the same ion, these troughs usually strengthen or weaken together. We explore this phenomenon for C iv, finding that 78% of high-velocity troughs vary in a ≈ 99 coordinated manner with the lowest-velocity trough. Coordinated trough variations are found even for velocity offsets as large as 15000–20000 km s−1. See Section 3.4.7. 11. We have examined if BAL EW variations on several timescales depend upon quasar properties, including LBol, LBol/LEdd, MBH, redshift, and radio loudness. Within the ranges of these properties spanned by our sample, we do not find any strong dependences. There may be a hint that C iv EW variations depend on LBol and/or LBol/LEdd. See Section 3.4.8. Considering the observational results above, we discuss implications for quasar out- flows. The main implications are the following: 1. The high observed frequency of BAL variability on multi-year timescales is generally supportive of models where most BAL absorption arises at radii of 10–1000 light days. However, a significant minority of BAL troughs are remarkably stable and may be associated with absorption on larger scales. See Section 3.5.1. 2. Our measurements of EW variations as a function of timescale can be used to infer an (EW-dependent) average lifetime for a BAL trough along our line-of-sight of tBAL = +2100 h i 1600−900 yr. This average lifetime is long compared to the orbital time of the accretion disk at the wind-launching or BLR radius. See Section 3.5.2. 3. Comparison of the ∆EW and ∆EW/ EW distributions with previously identified h i examples of BAL disappearance indicates that BAL disappearance is an extreme type of general BAL variability, rather than a qualitatively distinct phenomenon. The same appears to apply for BAL emergence. See Section 3.5.3. 4. We examine the extent to which a simple one-dimensional random-walk model can ex- plain the evolution of BAL-trough EWs. The binomial distribution resulting from this model can acceptably fit the ∆EW distribution for C iv BAL troughs on timescales of 2.0–2.5 yr, where we have the best trough statistics. The best-fitting model has a step size of 1.65 A˚ and six random-walk steps, corresponding to a rest-frame step timescale of 137 days. The small derived number of steps is consistent with the ≈ non-Gaussian nature of the ∆EW distribution, and we derive consistent random-walk parameters when considering the ∆EW distribution for Si iv troughs. A simulation of randomly walking trough EWs shows the best-fitting model derived for timescales of 2.0–2.5 yr also acceptably represents EW variations over a wider range of timescales. See Section 3.5.4. 5. Coordinated trough variability for BAL quasars showing multiple troughs implies the existence of a coordinating mechanism capable of affecting outflow streamlines spread over a wide range of radii. The mechanism may be changes in the shielding gas that lead to changes in the level of EUV/X-ray radiation reaching the streamlines. 100 Such changes in shielding gas may be a driver of BAL variability more generally. See Section 3.5.5. The ongoing BOSS ancillary project on quasar-wind variability continues to enlarge the sample of BAL quasars with high-quality spectra spanning multiple years, and we hope to obtain an additional 1–2 epochs of observation for the associated BAL quasars as part of the Time Domain Spectroscopic Survey (TDSS) of the SDSS-IV.4 These data should enable multiple further studies of BAL variability, and here we highlight a few selected ways that the main results found above might be advanced. First, in this paper we have focused upon variability of the relatively high-ionization transitions of C iv and Si iv, and it would be worthwhile to extend large-sample analyses to other transitions including Mg ii, Al iii, Fe ii, and Fe iii (e.g., Hall et al. 2011; Vivek et al. 2012). The outflow structure along the line-of-sight may differ significantly when such lower ionization transitions are present. Second, additional data can improve constraints upon the fraction of BALs show- ing variability, BAL lifetimes, and our simple random-walk model for BAL EW variability. Ideally, one would also like to test our measurements of BAL variability against specific predictions made by advancing models of quasar winds. Third, our results on coordinated trough variability for BAL quasars showing multiple troughs indicate that large-scale inves- tigations of BAL vs. continuum variability are important to pursue. Continuum variability investigations in the rest-frame UV should be possible once the BOSS spectrophotometric calibration is improved, and limited X-ray continuum monitoring should be possible with targeted observations and the eROSITA all-sky survey (e.g., Merloni et al. 2012). Fourth, a large-scale investigation of BAL vs. emission-line variability should give unique insights into the quasar-wind contribution to high-ionization line production. Finally, additional observational epochs will allow a large-scale search for BAL acceleration events which have generally proved elusive to date (e.g., Gibson et al. 2008, 2010; Capellupo et al. 2012). 4The current planning for SDSS-IV is briefly described at http://www.sdss3.org/future/ 101 Table 3.1. Sample-Based Studies of BAL Quasar Variability Reference #ofQuasars ∆t Range # of Epochs (yr) Barlow (1993) 23 0.2–1.2 2–6 Lundgren et al. (2007) 29 0.05–0.3 2 Gibson et al. (2008) 13 3.0–6.1 2 Gibson et al. (2010) 14 0.04–6.8 2–4 Capellupo et al. (2011, 2012, 2013) 24 0.02–8.7 2–13 Vivek et al. (2012)a 5 0.01–5 4–14 Haggard et al. (2012) 17 0.001–0.9 6 Filiz Ak et al. (2012)b 19 1.1–3.9 2–4 Welling et al. (2013)c 46 0.2–16.4 2–6 This study 291 0.0006–3.7 2–12 Full BOSS Ancillary 2105 0.0006–6 2–12 aFe low-ionization BAL quasars bQuasars with disappearing BAL troughs cRadio-loud BAL quasars 102 Table 3.2: C iv BAL Troughs iv a b C Quasar Name RA Dec z σz LoBAL Timescale Plate[1] MJD[1] TroughID J2000 J2000 Flag Flag ··· C1 J001502.26+001212.4 3.75943 0.20346 2.8525 0.00055 0 1 389 51795 C2 J001502.26+001212.4 3.75943 0.20346 2.8525 0.00055 0 1 389 51795 C3 J003135.57+003421.2 7.89823 0.57257 2.2364 0.00026 0 1 689 52262 C4 J003135.57+003421.2 7.89823 0.57257 2.2364 0.00026 0 1 689 52262 C5 J003517.95+004333.7 8.82481 0.72604 2.9169 0.00055 0 1 1086 52525 Fiber[1] SN1700[1] Plate[2] MJD[2] Fiber[2] SN1700[2] ∆t vmax vmin vcent[1] vcent[2] (days) (km s−1) (kms−1) (kms−1) (kms−1) ··· 465 10.62 4218 55479 818 21.43 956.253 23238.4 19878.0 21752.0 21650.5 465 10.62 4218 55479 818 21.43 956.253 −10478.3 − 5841.4 − 8427.4 − 8285.5 502 15.06 3587 55182 570 24.52 902.230 −21045.9 −18737.9 −19810.5 −19820.2 502 15.06 3587 55182 570 24.52 902.230 −12539.9 − 4858.5 − 9649.1 − 9361.5 481 12.27 3587 55182 722 14.60 678.344 −20339.7 −16881.9 −18683.1 −18626.6 − − − − Continued on Next Page... a1 for low-ionization BAL quasars, 0 otherwise b 1 for timescales of more than 1 yr (∆tmin,1), 0 otherwise (∆tmin) 103 Table 3.2–Continued ∆EW iv ∆EW C dBAL[1] σdBAL [1] dBAL[2] σdBAL [2] EW[1] σEW[1] EW[2] σEW[2] ∆EW σ∆EW EW σ h i hEWi ··· Trough ID (A˚ ) (A˚ ) (A˚ ) (A˚ ) (A˚ ) (A˚ ) C1 0.21 0.023 0.13 0.006 3.71 0.59 2.4 0.16 1.34 0.61 0.442 0.2147 C2 0.46 0.034 0.44 0.030 11.51 0.56 10.3 0.13 −1.19 0.58 −0.109 0.0611 C3 0.14 0.024 0.21 0.022 1.65 0.13 2.5 0.07− 0.87 0.15− 0.415 0.1025 C4 0.69 0.025 0.73 0.025 26.86 0.20 28.5 0.10 1.61 0.22 0.058 0.0109 C5 0.23 0.022 0.22 0.012 3.99 0.37 3.9 0.25 0.05 0.44 0.012 0.1549 − − N c ∆v f Corresp. Si ivd log L σ log M σ log LBol M R VR VR ∆v Bol log LBol BH log MBH LEdd i −1 −1 −1 (kmsP ) TroughID (ergs ) (ergs ) (M⊙) (M⊙) 1 274.7 0.082 46.970 0.02 9.34 0.12 0.468 27.68 1.0 0 0.0 0.000··· S1 46.970 0.02 9.34 0.12 −0.468 −27.68 −1.0 0 0.0 0.000 47.040 0.01 9.38 0.12 −0.441 −27.65 −1.0 2 1655.7 0.216··· S2 47.040 0.01 9.38 0.12 −0.441 −27.65 −1.0 0 0.0 0.000 46.880 0.01 9.86 0.18 −1.076 −27.61 −1.0 ··· − − − cNumber of variable regions found in the BAL trough dCorresponding Si iv BAL trough ID as given in Table ?? Note. — Throughout this table [1] indicates the first-epoch spectra and [2] indicates the second-epoch spectra. (This table is available in its entirety in a machine-readable form in the online journal. A portion is shown here for guidance regarding its form and content.) 104 Table 3.3: Si iv BAL Troughs iv a b Si Quasar Name RA Dec z σz LoBAL Flag Timescale Flag Plate[1] MJD[1] TroughID J2000 J2000 ··· S1 J001502.26+001212.4 3.75943 0.20346 2.8525 0.00055 0 1 389 51795 S2 J003135.57+003421.2 7.89821 0.57260 2.2364 0.00026 0 1 689 52262 S3 J003517.95+004333.7 8.82481 0.72604 2.9169 0.00055 0 1 1086 52525 S4 J004613.54+010425.7 11.55643 1.07381 2.1492 0.00028 0 1 691 52199 S5 J005419.99+002727.9 13.58329 0.45776 2.5143 0.00025 0 1 394 51913 Fiber[1] SN1700[1] Plate[2] MJD[2] Fiber[2] SN1700[2] ∆t vmax vmin vcent[1] vcent[2] (days) (km s−1) (kms−1) (kms−1) (kms−1) ··· 465 10.62 4218 55479 818 21.43 956.2525 12074.1 5872.0 8972.6 8908.4 502 15.06 3587 55182 570 24.52 902.2298 −12521.9 −4933.6 −8905.7 −8831.1 481 12.27 3587 55182 722 14.60 678.3438 − 5224.5 −3000.0 −4286.6 −4296.7 460 23.70 3589 55186 866 28.94 948.4810 −20051.0 −9612.2 −14741.9 −14841.3 511 31.12 4224 55481 726 25.32 1015.2949 −11686.2 −5045.8 − 8284.2 − 8293.7 − − − − Continued on Next Page... a1 for low-ionization BAL quasars, 0 otherwise b 1 for timescales of more than 1 yr (∆tmin,1), 0 otherwise (∆tmin) 105 Table 3.3–Continued ∆EW iv ∆EW Si dBAL[1] σdBAL [1] dBAL[2] σdBAL [2] EW[1] σEW[1] EW[2] σEW[2] ∆EW σ∆EW EW σ h i hEWi ··· Trough ID (A˚ ) (A˚ ) (A˚ ) (A˚ ) (A˚ ) (A˚ ) S1 0.221 0.0159 0.139 0.0059 6.41 0.68 3.97 0.12 2.44 0.69 0.4707 0.1292 1 S2 0.311 0.0158 0.361 0.0175 10.84 0.24 12.60 0.12− 1.76 0.27− 0.1499 0.0315 1 S3 0.439 0.0456 0.432 0.0437 4.76 0.21 4.69 0.13 0.07 0.25 0.0144 0.0731 0 S4 0.223 0.0058 0.171 0.0082 10.46 0.29 8.16 0.17 −2.30 0.34 −0.2467 0.0484 3 S5 0.200 0.0087 0.174 0.0107 6.11 0.16 5.33 0.12 −0.78 0.20 −0.1371 0.0475 3 − − c ivd LBol NVR ∆vVR f∆v Corresp. C log LBol σlog LBol log MBH σlog MBH log LEdd Mi R −1 −1 −1 (kmsP ) TroughID (ergs ) (ergs ) (M⊙) (M⊙) 1 276.1 0.045 C2 46.97 0.020 9.342 0.121 0.468 27.682 1 1 345.1 0.046 C4 47.04 0.007 9.380 0.117 −0.441 −27.654 −1 0 00 46.88 0.014 9.855 0.182 −1.076 −27.610 −1 3 1858.4 0.178 ··· 47.08 0.009 9.220 0.220 −0.245 −28.404 11.4− 3 1311.2 0.198 C10··· 47.07 0.005 9.652 0.043 −0.681 −28.209 1 − − − cNumber of variable regions found in the BAL trough dCorresponding C iv BAL trough ID as given in Table ??. As explained in Section 3.4.6, we search for accompanying C iv BAL iv −1 iv troughs for Si troughs with 3000 > vcent > 13000 km s . We found that five Si troughs, S3, S13, S201, S202, and S237, have an accompanying C iv trough− that does not− satisfy our BAL-trough identification criteria. Note. — Throughout this table [1] indicates the first-epoch spectra and [2] indicates the second-epoch spectra. (This table is available in its entirety in a machine-readable form in the online journal. A portion is shown here for guidance regarding its form and content.) 106 Table 3.4: C iv BAL-Trough Variable Regions a C iv Quasar Name MJD[1] MJD[2] C iv vmax vmin vmax,VR vmin,VR ∆vVR vnrt VR ID Trough ID (kms−1) (kms−1) (kms−1) (kms−1) (kms−1) CV1 J001502.26+001212.4 51795 55479 C1 23238.4 19878.0 23160.0 22885.4 274.7 0.85 CV2 J003135.57+003421.2 52262 55182 C4 −12539.9 − 4858.5 −11608.3 −10642.5 965.8 −0.51 CV3 J003135.57+003421.2 52262 55182 C4 −12539.9 −4858.5 −10504.5 − 9814.5 690.0 −0.18 CV4 J004323.43 001552.4 52261 55184 C6 −21405.5 −10137.6 −16564.3 −15186.6 1377.7− 0 CV5 J004323.43−001552.4 52261 55184 C6 −21405.5 −10137.6 −15048.8 −12153.5 2895.3 0.40 − − − − − Note. — Throughout this table [1] indicates the first-epoch spectra and [2] indicates the second-epoch spectra. (This table is available in its entirety in a machine-readable form in the online journal. A portion is shown here for guidance regarding its form and content.) 107 Table 3.5: Si iv BAL-Trough Variable Regions a Si iv Quasar Name MJD[1] MJD[2] Si iv vmax vmin vmax,VR vmin,VR ∆vVR vnrt VR ID Trough ID (kms−1) (kms−1) (kms−1) (kms−1) (kms−1) SV1 J001502.26+001212.4 51795 55479 S1 12074.1 5872.0 8365.8 8089.7 276.1 0.24 SV2 J003135.57+003421.2 52262 55182 S2 −12521.9 −4933.6 −9137.9 −8792.8 345.1− 0.02 SV3 J004613.54+010425.7 52199 55186 S4 −20051.0 −9612.2 −19915.7 −18952.7 963.0 0.91 SV4 J004613.54+010425.7 52199 55186 S4 −20051.0 −9612.2 −18126.9 −17507.4 619.5 0.60 SV5 J004613.54+010425.7 52199 55186 S4 −20051.0 −9612.2 −11649.9 −11374.0 275.9 0.61 − − − − − Note. — Throughout this table [1] indicates the first-epoch spectra and [2] indicates the second-epoch spectra. (This table is available in its entirety in a machine-readable form in the online journal. A portion is shown here for guidance regarding its form and content.) 108 Table 3.6. Variability Comparison Between the Lowest Velocity BAL Trough and Additional BAL Troughs Number of Additional C iv Troughs ∆ EW < 0 ∆ EW > 0 ∆EWlow vel < 0 57 13 ∆EWlow vel > 0 17 50 109 Table 3.7. Spearman Rank Correlation Test Probabilities for Quasar Properties vs. BAL-Trough EW Variations C iv Si iv < 1 yr 1–2.5 yr > 2.5 yr < 1 yr 1–2.5 yr > 2.5 yr LBol ∆ EW 78.4 0.2 5.2 65.3 5.5 79.8 | | ∆ EW 31.5 83.2 67.1 44.3 90.6 11.1 | hEW i | LBol/LEdd ∆ EW 54.6 0.1 35.1 65.6 81.5 96.2 | | ∆ EW 26.3 2.4 25.1 39.5 92.5 25.4 | hEW i | MBH ∆ EW 32.9 11.4 66.4 56.1 85.1 66.6 | | ∆ EW 45.3 1.2 32.7 38.5 85.1 7.8 | hEW i | z ∆ EW 70.4 1.2 12.5 49.9 31.1 6.8 | | ∆ EW 8.4 3.4 48.3 32.9 16.9 30.9 | hEW i | Numberofdatapoints 82 220 126 56 124 55 Note. — Numbers in this table show the Spearman rank correlation test probabilities as percentages. 110 Chapter 4 The Dependence of C iv Broad Absorption Line Properties on Accompanying Si iv and Al iii Absorption: Relating Quasar-Wind Ionization Levels, Kinematics, and Column Densities 4.1 Introduction Broad Absorption Lines (BALs) in quasar spectra are seen as a result of high-velocity outflows. The most commonly used empirical definition of BALs requires an absorption feature to have at least a 2000 km s−1 width at 10% under the continuum level (e.g., Weymann et al. 1991). BAL quasars exhibit such broad absorption troughs in a wide variety of species in their rest-frame ultraviolet spectra, such as P v λλ1118, 1128 A,˚ Lyα λ1216 A,˚ N v λλ1239, 1243 A,˚ Si iv λλ1394, 1403 A,˚ C iv λλ1548, 1551 A,˚ Al ii λ1671 A,˚ Al iii λλ1855, 1863 A,˚ and Mg ii λλ2797, 2804 A.˚ BAL quasars are classified into three groups based on the observed transitions in their spectra. The majority of BAL quasars exhibit absorption from only high-ionization transitions such as N v, Si iv, and C iv (HiBALs, e.g., Weymann et al. 1991). Approx- imately 10% of BAL quasars in optically selected samples also exhibit absorption from low-ionization transitions such as Al ii, Al iii, and Mg ii (LoBALs, e.g., Voit et al. 1993; Gibson et al. 2009). Only 1% of BAL quasars show absorption from excited states of Fe ii ≈ and/or Fe iii in addition to the high and low-ionization transitions listed above (FeLoB- ALs, e.g., Becker et al. 2000; Hall et al. 2002). The existence of these groups indicates that quasar outflows can have a wide range of ionization states. It has been argued that the presence of the low-ionization lines is not the only difference between these groups (e.g., Boroson & Meyers 1992; Turnshek et al. 1994; Zhang et al. 2010). The generally weak [O iii] emission and strong reddening of LoBALs suggest that LoBALs tend to be surrounded by dust and gas that has a larger global covering factor compared to HiBALs. The details of the structure and geometry of quasar outflows remain unclear. A commonly adopted and well-developed model suggests that many BALs are formed in an equatorial wind that is launched from the accretion disk at 1016–1017 cm from the central ≈ supermassive black hole (SMBH) and is driven by radiation pressure (e.g., Murray et al. 111 1995; Proga et al. 2000; Higginbottom et al. 2013). This disk-wind model successfully explains several important observational facts about BAL quasars, such as the presence of absorption from both high and low-ionization transitions despite the luminous ionizing radiation from the central source, and the large range of outflow velocities that reach from the systemic velocity up to 0.1c. Numerical hydrodynamical simulations of the disk-wind model provide detailed predictions of the structure and dynamics of quasar outflows (e.g., Proga et al. 2000). Previous studies have shown that some BAL troughs are much more optically thick than they appear and that the depths of such troughs only mildly depend on column den- sity. Several pieces of evidence for this line-saturation interpretation have been presented, such as P v BALs, depth differences in unblended doublet lines, and “flat-bottom” BAL profiles (e.g., Arav 1997; Hamann 1998; Arav et al. 1999a,b, 2001; Leighly et al. 2009; Borguet et al. 2012). For instance, Hamann (1998) studied spectra of the BAL quasar PG 1254+047 which possesses relatively strong P v absorption at velocities corresponding to non-black strong C iv and S iv BAL troughs. The existence of P v absorption was taken as evidence for saturated C iv absorption since phosphorus is expected to be 1000 times ∼ less abundant than carbon (based on solar abundances). As another example, Arav et al. (1999b) argued that the depth differences between the unblended Si iv doublet lines of the quasar FIRST J1603+3002 arise as a result of velocity-dependent partial coverage. Calcu- lating the optical depths, they found that the C iv and Si iv absorption lines are saturated. Such studies have suggested that the non-black nature of these saturated lines arises due to partial coverage of the emission source along the line-of-sight; BAL troughs do not reach zero intensity due to photons from the emission source that are not absorbed and/or are scattered into the observer’s line-of-sight. Supporting this argument, spectropolarimetric observations of BAL quasars have shown that the fractional contribution from scattered emission often increases at the wavelengths where BAL troughs are found, indicating an excess of scattered light relative to direct light (e.g., Ogle et al. 1999). These lines of obser- vational evidence indicate that some BAL quasars have highly saturated BAL troughs and that the depths of such troughs are largely set by the line-of-sight covering factor rather than column density. However, detailed studies of line saturation are only available for a limited number of objects and have often focused on deep C iv BALs. It is possible that some BALs might be weak simply because the column density along the line-of-sight is small. Recent sample-based investigations of BAL variability have brought new insights about the structure, dynamics, and evolution of quasar outflows showing that variability is common for most BAL troughs on multi-year timescales (e.g., Lundgren et al. 2007; Gibson et al. 2008, 2010; Capellupo et al. 2011, 2012; Filiz Ak et al. 2012, 2013; Vivek et al. 2012; Wildy et al. 2013). These studies have revealed that the fractional variation of BAL troughs in lower ionization transitions (such as Si iv) is generally stronger than that in C iv (e.g., Capellupo et al. 2012; Vivek et al. 2012; Filiz Ak et al. 2013). A recent study by Filiz Ak et al. (2013) presented a detailed investigation of the variability of C iv and Si iv 112 BALs on multi-year timescales in a large quasar sample assessing variation characteristics and the lifetimes of BAL troughs. This study found coordinated trough variability for BAL quasars showing multiple C iv troughs; they suggested that global changes in ionization level are the most straightforward mechanism for explaining such coordinated variability of multiple C iv troughs at different velocities. This mechanism would require at least some BAL troughs not to be highly saturated, as highly saturated troughs should not be responsive to the expected changes in ionization level. The available analytic calculations and numerical simulations of quasar disk winds predict the ionization level, kinematics, and column density of the outflowing gas along possible lines-of-sight to the relevant emission region (e.g., Murray et al. 1995; Proga et al. 2000; Higginbottom et al. 2013). These three quantities are expected generally to show correlated changes as the line-of-sight is varied. Thus, we expect correlated object-to-object changes of resulting observable phenomena such as the BAL transitions present, BAL- profile shapes, and BAL variability. In this paper, we aim to investigate systematically and quantify such correlated object-to-object changes for a large sample of BAL quasars with uniform high-quality measurements from the Sloan Digital Sky Survey (SDSS, York et al. 2000). Utilization of a large sample is important to overcome object-to-object scatter associated with, e.g., time-variable wind inhomogeneities. To probe correlated object-to-object changes of ionization level, kinematics, and column density, we require a basic means of identifying lines-of-sight with different average ionization levels. We accomplish this using the strong BAL transitions of C iv, Si iv, and Al iii. These three transitions span a significant range of ionization potential (with creation ionization potentials of 47.9, 33.5, and 18.8 eV, respectively, e.g., Hall et al. 2002), and their BAL regions can all be measured simultaneously in SDSS spectra of quasars with redshift 1.9 4.2 Observations, Sample Selection, and Data Preparation We utilize spectroscopic observations from the Sloan Digital Sky Survey-I/II (here- after SDSS, York et al. 2000) and the Baryon Oscillation Spectroscopic Survey of SDSS-III (hereafter BOSS, Eisenstein et al. 2011; Dawson et al. 2013). SDSS is a large multi-filter imaging and spectroscopic survey using a dedicated 2.5-m optical telescope (Gunn et al. 1998, 2006) at Apache Point Observatory in New Mexico. During its first phase of op- erations, 2000–2005, the SDSS imaged more than 8000 deg2 of the sky in five optical bandpasses, and it obtained spectra of galaxies and quasars. The SDSS spectral coverage was continuous from 3800 A˚ to 9200 A˚ at a resolution of 1800–2200 (e.g., York et al. 2000). In 2005 the survey entered a new phase, the SDSS-II, expanding its spectroscopic samples to over 800,000 galaxies and 100,000 quasars. The BOSS, part of the third phase of SDSS operations, is acquiring spectra for approximately 1.5 million luminous galaxies and 160,000 quasars (e.g., Anderson et al. 2012; Ross et al. 2012). The BOSS survey started operating in mid-2008 and is planned to continue observations until the end of June 2014. The BOSS quasar survey provides an outstanding opportunity for investigating intrinsic UV absorption in quasars, owing to its focus upon selection at z > 2.1 which shifts the important C iv and Si iv transitions well into its spectral coverage (Smee et al. 2013). The BOSS spectral coverage is continuous from 3600 A˚ to 10,000 A˚ at a resolution of 1300–3000 (e.g., Dawson et al. 2013). An ancillary BOSS project aims to investigate the dynamics of quasar winds over multi-year timescales utilizing second-epoch spectra for 2005 BAL quasars originally iden- tified in the SDSS-I/II spectroscopy by Gibson et al. (2009). These 2005 quasars were −1 initially selected following i< 19.3, 0.48 1. We select quasars with 1.9 2. We select quasars that have a C iv balnicity index between 3000 and 20,000 km s−1, 20 − − BI3 , greater than 0 for both the SDSS and BOSS spectra; implementing this re- quirement for both SDSS and BOSS spectra avoids biases that could arise from non-uniform SDSS vs. BOSS BAL-identification thresholds. Thus the quasars with disappearing BAL troughs in Filiz Ak et al. (2012) are not included in this study. 20 We define BI3 using the following equation: −20000 20 f(v) BI3 1 Cdv. (4.1) ≡ Z−3000 − 0.9 Similar to the original BI definition (Weymann et al. 1991), in this equation f(v) is the normalized flux density as a function of velocity, v. C is a constant which is equal to 1.0 only when a trough is wider than 2000 km s−1; it is otherwise 0.0. Following the original BI definition, the minimum red-edge velocity limit is chosen to minimize confusion between C iv BALs and the C iv emission line. 3. We select only radio-quiet quasars by requiring the radio-loudness parameter, R, to be less than 10; we utilize R parameters from Shen et al. (2011). Considering that radio- loud quasars are a minority part of the quasar population and may have different BAL properties than radio-quiet quasars (e.g., Becker et al. 1995, 2000; Brotherton et al. 1998), implementing this criterion avoids possible confusion associated with the presence of an additional radio-loud population. 20 Differing from the original BI definition of Weymann et al. (1991), our adopted BI3 definition for this study (see Equation 4.1) limits the maximum blue-edge velocity of a BAL-trough region at 20,000 km s−1, where it is 25,000 km s−1 in the original definition. Given that this study is focused on the characteristics of and differences between C iv BAL troughs that are accompanied by Si iv and/or Al iii BALs in corresponding velocity ranges, we adjust the maximum blue-edge velocity limit of the C iv BAL-trough region considering the Si iv (1394, 1403 A)˚ and Al iii (1855, 1863 A)˚ BAL-trough regions. Both the Si iv and Al iii BAL-trough regions are occasionally contaminated by the emission lines of O i 1302 A˚ (at 21,800 km s−1 from Si iv emission), Si ii 1304 A˚ (at 21,300 km s−1 ≈− ≈− from Si iv emission), C ii 1334 A˚ (at 14,500 km s−1 from Si iv emission), Ni ii 1741 ≈ − and 1751 A˚ (at 20,200 and 18,400 km s−1 from Al iii emission), and Fe ii 1787 A˚ (at ≈− − 12,500 km s−1 from Al iii emission). Although these emission lines are usually weak, ≈ − a visual inspection showed that in some cases these features may bring an end to shallow BAL troughs. Moreover, low-velocity absorption from these other line species may lead to confusion in the detection of Si iv or Al iii absorption. Thus, we consider a BAL-trough region between 3000 and 20,000 km s−1 as a suitable balance between uncontaminated − − spectral regions and useful sample size. Both C ii and Fe ii have low ionization potentials (11.3 eV and 2.8 eV, respectively) and are rarely found in quasar spectra (e.g., Hall et al. 2002). 115 From the initial set of 2005 BAL quasars, the above criteria select 714 quasars that are observed by both the SDSS and BOSS (the main sample will be reduced to 671 quasars below via further considerations). The median SN1700 is 10.7 for SDSS and 17.4 for BOSS observations. In order to compare multi-epoch observations of our main sample, for each spectrum, we follow the basic spectral preparation procedures given in Section 3.1 of Filiz Ak et al. (2012). These procedures include Galactic extinction correction using the AV values from Schlegel et al. (1998), transforming from the observed frame to the rest frame using the redshift values from Hewett & Wild (2010), and removing pixels that are flagged by the SDSS and BOSS data-reduction pipelines (Bolton et al. 2012). As in Gibson et al. (2008, 2010) and Filiz Ak et al. (2012, 2013), we define relatively line-free (RLF) windows to reconstruct the underlying continuum. We fit the RLF windows of each spectrum with an intrinsically reddened power-law using a Small Magellanic Cloud type reddening model (e.g., Pei 1992). This fit is performed by running an iterative sigma- clipping algorithm where in each iteration we perform a non-linear least squares fit. We calculate the uncertainties on the continuum model using ∆χ2 confidence-region estimation as described in Filiz Ak et al. (2012). We do not model the emission lines. We follow the procedures of Filiz Ak et al. (2012, 2013) for error propagation of continuum uncertainties into the subsequent measurements. 4.3 Identification of BAL Troughs and Measurements 4.3.1 Identification of BAL Troughs We identify C iv BAL troughs in the spectra of the 714 quasars that satisfy our quasar selection criteria (see Section 4.2 using the definition in Equation 4.1 and following the BAL-trough identification algorithm for multi-epoch observations defined in Section 3.2 of Filiz Ak et al. 2013). By construction, each quasar in our main sample has at least one SDSS and at least one BOSS observation. However, 25% of the main-sample quasars ≈ have multiple SDSS and/or BOSS observations. These repeat observations sample similar timescales (considering just the multi-year timescales of primary interest here) and could cause some of our objects to be given inordinate weight by allowing the repeat examination of BAL troughs. In order to avoid such multi-counting biases in our analyses, we use only two-epoch spectra for each quasar in our main sample. We select the one SDSS-I/II spectrum and the one BOSS spectrum that have the highest SN1700. Using these two-epoch observations, we classify C iv BAL troughs into three groups (denoted with subscripts) as explained below: 1. C iv00 indicates C iv BAL troughs with no detection of BAL or mini-BAL troughs at corresponding velocities in the Si iv and Al iii absorption regions in both epochs. 116 iv iv iv 2. C S0 indicates C BAL troughs accompanied by a Si BAL trough in either epoch but with no detection of a BAL or mini-BAL trough at corresponding velocities in the Al iii BAL region.1 3. C ivSA indicates C iv BAL troughs accompanied by a Si iv BAL and also an Al iii BAL detected at corresponding velocities in either epoch. We did not find any examples of C iv BAL troughs accompanied by an Al iii BAL but iv iv iii no Si BAL (i.e., C 0A); lower ionization absorption (in this case, Al ) troughs are always accompanied by higher ionization (Si iv and C iv) troughs at the same velocities. As discussed in Section 4.1, these three C iv groups serve as a basic means for identifying lines-of-sight with different average ionization levels. We identify 43 quasars showing multiple C iv BAL troughs that are classed in differ- ent C iv groups (38 with C iv00 and C ivS0 troughs, and 5 with C ivS0 and C ivSA troughs). While the existence of such objects is entirely expected within accretion-disk wind models, practically they introduce complexity in connecting a given BAL trough to a given line-of- iv iv sight; e.g., C 00 troughs in quasars that also show C S0 troughs likely sample a different iv part of the outflow from C 00 troughs in quasars with no other BALs (see Section 4.5 for further discussion). To reduce complexity and avoid potentially mixing troughs in the same C iv group that sample different parts of the outflow, we exclude from our main sam- ple these 43 quasars. We have performed all analyses below also including such quasars, and the results do not change materially. After excluding these 43 quasars, there are 671 quasars in our main sample (see Table 4.1). We detect a total of 852 C iv BAL troughs in the spectra of these 671 main-sample quasars as our main-sample C iv troughs. We have identified 113 C iv00, 246 C ivS0, and 95 C ivSA BAL troughs in our main sample. We present our measurements for C iv00 troughs in Table 4.2, for C ivS0 and corresponding Si iv BAL troughs in Table 4.3, and for C ivSA and corresponding Si iv and Al iii BAL troughs in Table 4.4. In addition to these groups we have detected 273 C iv BAL troughs accompanied by a Si iv mini-BAL and 125 C iv BAL troughs accompanied by a Si iv BAL/mini-BAL and also an Al iii mini-BAL. We do not include these troughs in any of the groups defined above to consider only strong intrinsic absorption of the given transition; this approach should give the strongest distinction between groups. Figures 4.1, 4.2, and 4.3 show two-epoch observations for representative examples of C iv00,C ivS0, and C ivSA BAL troughs, respectively. We smoothed each spectrum for display purposes using a Savitzky-Golay algorithm (see Press et al. 1992) that performs a local linear regression for three consecutive points (see Filiz Ak et al. 2012). In these figures, we indicate C iv BAL troughs and their corresponding velocity ranges in the Si iv 1We define a mini-BAL using Equation 4.1 but adapting the constant, C, equal to one for absorption lines with 500 < ∆v < 2000 kms−1 and zero otherwise (e.g., Filiz Ak et al. 2013, and references therein). 117 and Al iii absorption regions. The SDSS identification, redshift, and timescale between the two epochs are given in each panel. In this study, we generally focus on the C iv, Si iv, and Al iii BAL regions. We do not analyze in detail the BAL regions of the other transitions listed in Section 4.1 for several reasons: (1) Due to the limited wavelength coverage of the observed spectra (3800–9200 A˚ for SDSS, and 3600–10000 A˚ for BOSS), only a small number of quasars have suitable wavelength coverage to investigate all of these transitions. (2) The strong Lyα and N v lines blend with each other and their blueshifted absorption lines blend with Lyα forest, where numerous absorption features from intervening gas are found. (3) A proper investigation of Mg ii BALs would require modeling and subtraction of broad Fe ii emission in addition to underlying continuum estimation. Moreover, only a small fraction of our targeted quasars have the wavelength coverage to investigate simultaneously C iv, Si iv, and Mg ii BAL troughs. 4.3.2 Measurements of BAL Troughs We measure the rest-frame timescales, ∆t, for our main sample; ∆t is between 0.85 and 4.13 yr with a mean of 2.55 yr. Given the established connection between BAL- variation strength and timescale (e.g., Gibson et al. 2010; Filiz Ak et al. 2013), we compared the ∆t distributions for C iv00,C ivS0, and C ivSA troughs. In Figure 4.4, we show the ∆t distributions for all C iv troughs in our main sample and for C iv00,C ivS0, and C ivSA troughs. Given their definition, the sum of the total number of C iv00,C ivS0, and C ivSA troughs is not equal to the total number of C iv troughs in our sample. We compared these ∆t distributions using two-sample Anderson-Darling (AD) tests (see Press et al. 1992). The test results show no significant differences between the distributions. Considering that BAL troughs are sometimes isolated and occasionally appear in complexes in which single troughs may split or adjacent troughs may merge over time, we define minimum and maximum velocities for each BAL trough following the BAL-trough identification algorithm for multi-epoch observations described in Section 3.2 of Filiz Ak et al. (2013). We set vmin to be the minimum red-edge velocity and vmax to be the maximum blue-edge velocity of the associated absorption complex in all available epochs. We measure the rest-frame EW of each BAL trough in each epoch. To calcu- late uncertainties on EWs, we propagate observational errors for each contributing pixel and continuum-estimation errors (see Filiz Ak et al. 2012, 2013) using Equations 1 and 2 of Kaspi et al. (2002). We calculate EW variations, ∆EW, fractional EW variations, ∆EW/ EW , and uncertainties on these quantities, σ and σ , following Equa- h i ∆EW ∆EW/hEWi tions 3 and 4 of Filiz Ak et al. (2013), respectively. In this study, positive values of ∆EW and ∆EW/ EW indicate strengthening troughs and negative values indicate weakening h i troughs. In addition, we measure the average depth, dBAL, for each trough by calculating the mean distance of each contributing data point from the normalized continuum level. We calculate a BAL-trough velocity width, ∆v, and a weighted centroid velocity, vcent, 118 Velocity −20 −15 −10 −5 0 −20 −15−10 −5 0 −20 −15 −10 −5 0 λ 2.5 SDSS J103407.74+452830.6 z =2.45 2.0 ∆ t =2.68 yr 1.5 1.0 0.5 SDSS Si IV C IV Al III BOSS Normalized Flux Density F 2 σ 0 N −2 1300 1400 1500 1600 1700 1800 1900 Rest Wavelength Velocity −20 −15 −10 −5 0 −20 −15−10 −5 0 −20 −15 −10 −5 0 λ SDSS J105505.30+415848.6 1.5 z =2.66 ∆ t =1.94 yr 1.0 0.5 SDSS Si IV C IV Al III BOSS Normalized Flux Density F 2 σ 0 N −2 1300 1400 1500 1600 1700 1800 1900 Rest Wavelength Velocity −20 −15 −10 −5 0 −20 −15−10 −5 0 −20 −15 −10 −5 0 λ 2.0 SDSS J164551.49+372653.7 z =2.09 1.5 ∆ t =3.26 yr 1.0 0.5 SDSS Si IV C IV Al III BOSS Normalized Flux Density F 5 σ 0 N −5 1300 1400 1500 1600 1700 1800 1900 Rest Wavelength Fig. 4.1: Two-epoch spectra of quasars with C iv00 BAL troughs (i.e., C iv BAL troughs with no detection of BAL or mini-BAL troughs at corresponding velocities in the Si iv and Al iii absorption regions) from SDSS (red) and BOSS (black). The x-axes show both the rest-frame wavelength (bottom, in A˚ ) and the blueshift velocity from the Si iv,C iv, and Al iii emission lines (top, in 103 kms−1). The y-axes show flux densities normalized iv by the fitted continuum model (Fλ). The horizontal solid-blue bars designate C BAL troughs, and the dashed blue bars indicate corresponding velocities in the Si iv and Al iii BAL regions. The lower section of each panel shows deviations between SDSS and BOSS observations for each 4 A˚ pixel in units of σ, N , and the dashed-red lines show 1σ ≈ σ ± levels. 119 Velocity −20 −15 −10 −5 0 −20 −15−10 −5 0 −20 −15 −10 −5 0 λ SDSS J083848.67+411703.9 2.0 z =2.85 ∆ t =2.59 yr 1.5 1.0 0.5 SDSS Si IV C IV Al III BOSS Normalized Flux Density F 2 σ 0 N −2 1300 1400 1500 1600 1700 1800 1900 Rest Wavelength Velocity −20 −15 −10 −5 0 −20 −15−10 −5 0 −20 −15 −10 −5 0 λ SDSS J100249.66+393211.0 1.5 z =2.26 ∆ t =2.18 yr 1.0 0.5 SDSS Si IV C IV Al III BOSS Normalized Flux Density F 5 σ 0 N −5 1300 1400 1500 1600 1700 1800 1900 Rest Wavelength Velocity −20 −15 −10 −5 0 −20 −15−10 −5 0 −20 −15 −10 −5 0 λ SDSS J223328.90+000104.3 z =2.09 1.5 ∆ t =2.88 yr 1.0 0.5 SDSS Si IV C IV Al III BOSS Normalized Flux Density F 5 σ 0 N −5 1300 1400 1500 1600 1700 1800 1900 Rest Wavelength Fig. 4.2: Same as Figure 4.1 but for C ivS0 BAL troughs (i.e., C iv BAL troughs accom- panied by a Si iv BAL trough but with no detection of a BAL or mini-BAL trough at corresponding velocities in the Al iii BAL region). 120 Velocity −20 −15 −10 −5 0 −20 −15−10 −5 0 −20 −15 −10 −5 0 λ SDSS J015921.53+141043.1 2.0 z =3.1 ∆ t =2.47 yr 1.5 1.0 0.5 SDSS Si IV C IV Al III BOSS Normalized Flux Density F 4 2 σ 0 N −2 −4 1300 1400 1500 1600 1700 1800 1900 Rest Wavelength Velocity −20 −15 −10 −5 0 −20 −15−10 −5 0 −20 −15 −10 −5 0 λ 2.0 SDSS J113527.26+385744.1 z =1.95 1.5 ∆ t =2.05 yr 1.0 0.5 SDSS Si IV C IV Al III BOSS Normalized Flux Density F 4 2 σ 0 N −2 −4 1300 1400 1500 1600 1700 1800 1900 Rest Wavelength Velocity −20 −15 −10 −5 0 −20 −15−10 −5 0 −20 −15 −10 −5 0 λ SDSS J150503.46−022324.4 1.5 z =2.09 ∆ t =2.57 yr 1.0 0.5 SDSS Si IV C IV Al III BOSS Normalized Flux Density F 5 σ 0 N −5 1300 1400 1500 1600 1700 1800 1900 Rest Wavelength Fig. 4.3: Same as Figure 4.1 but for C ivSA BAL troughs (i.e., C iv BAL troughs accom- panied by a Si iv BAL trough and also an Al iii BAL trough). 121 Table 4.1: Main-Sample BAL Quasars QuasarID QuasarName RA Dec z σz SDSS J2000 J2000 ··· Q1 J000119.64+154828.8 0.33184 15.80800 1.921029 0.000494 Q2 J001025.90+005447.6 2.60796 0.91324 2.859854 0.000323 Q3 J001502.26+001212.4 3.75943 0.20346 2.852539 0.000552 Q4 J003135.57+003421.2 7.89823 0.57257 2.236426 0.000255 Q5 J003312.25+155442.4 8.30105 15.91178 1.955444 0.000617 Q6 J003517.95+004333.7 8.82481 0.72604 2.916893 0.000545 Q7 J003551.98+005726.4 8.96660 0.95734 1.905903 0.000576 Q8 J003832.26+152515.5 9.63446 15.42100 2.448965 0.000500 Q9 J004732.73+002111.3 11.88639 0.35315 2.873223 0.000340 Q10 J005215.64+003120.5 13.06520 0.52236 2.792256 0.000321 Plate[1] MJD[1] Fiber[1] SN1700[1] Plate[2] MJD[2] Fiber[2] SN1700[2] ∆t (days) (days) (years) 750 52235 566 8.201 6172 56269 318 18.296 3.781 389 51795 332 6.379 4218 55479 592 20.610 2.613 389 51795 465 10.623 4218 55479 818 21.430 2.618 689 52262 502 15.060 3587 55182 570 24.501 2.470 417 51821 576 8.304 6192 56269 184 15.174 4.121 1086 52525 481 12.271 3588 55184 512 14.787 1.859 392 51793 449 8.931 3588 55184 552 24.975 3.195 418 51817 483 6.585 6197 56191 760 12.239 3.472 691 52199 559 19.051 4223 55451 724 26.027 2.299 084 52591 516 9.739 4223 55451 968 11.653 2.065 Notes. Throughout this table [1] indicates the first-epoch spectra and [2] indicates the second-epoch spectra. (This table is available in its entirety in a machine-readable form in the online journal. A portion is shown here for guidance regarding its form and content.) 122 Table 4.2: C iv00 Troughs Quasar C iv v v ∆v v [1] v [2] d [1] σ [1] 00 max min cent cent BAL dBAL ··· ID TroughID (kms−1) (kms−1) (kms−1) (kms−1) (kms−1) Q29 C00-1 10957.2 8528.4 2428.8 9712.9 9676.2 0.200 0.022 Q29 C00-2 − 7593.0 −4686.9 2906.1 −6046.1 −6072.0 0.339 0.024 Q34 C00-3 −20000.0 −17884.3 2115.7 −18892.9 −18901.7 0.124 0.009 Q34 C00-4 −17862.3 −12780.2 5082.1 −15274.8 −15295.9 0.130 0.009 Q40 C00-5 −15763.9 − 9381.9 6382.0 −12610.5 −12497.6 0.307 0.017 Q41 C00-6 −20000.0 −16100.5 3899.5 −17974.4 −17956.4 0.348 0.020 Q47 C00-7 −19397.8 −11712.7 7685.1 −15810.3 −15687.3 0.368 0.015 Q47 C00-8 −11537.6 − 8089.2 3448.4 − 9811.0 − 9823.4 0.329 0.023 Q51 C00-9 − 9963.7 −7962.3 2001.5 −8946.8 −8954.2 0.144 0.021 Q57 C00-10 −20000.0 −4959.0 15041.0 −12291.8 −12402.4 0.189 0.006 − − − − ∆EW ∆EW dBAL[2] σdBAL [2] EW[1] σEW[1] EW[2] σEW[2] ∆EW σ∆EW σ hEWi hEWi (A)˚ (A)˚ (A)˚ (A)˚ (A)˚ (A)˚ 0.147 0.015 2.37 0.19 1.65 0.18 0.72 0.26 0.36 0.18 0.311 0.020 5.10 0.20 4.71 0.17 −0.39 0.26 −0.08 0.08 0.133 0.005 1.27 0.16 1.42 0.05− 0.15 0.17− 0.11 0.16 0.145 0.005 3.18 0.25 3.67 0.08 0.49 0.26 0.14 0.10 0.294 0.013 9.49 0.50 9.37 0.21 0.12 0.54 0.01 0.07 0.279 0.015 6.76 0.22 5.45 0.18 −1.30 0.28 −0.21 0.06 0.118 0.010 13.77 0.61 3.70 0.24 −10.08 0.66 −1.15 0.07 0.189 0.016 5.68 0.35 3.21 0.13 − 2.47 0.37 −0.56 0.09 0.394 0.036 1.23 0.26 4.10 0.17− 2.87 0.31− 1.08 0.18 0.207 0.004 13.86 0.83 15.56 0.62 1.70 1.03 0.12 0.10 Notes. Throughout this table [1] indicates the first-epoch spectra and [2] indicates the second-epoch spectra. (This table is available in its entirety in a machine-readable form in the online journal. A portion is shown here for guidance regarding its form and content.) 123 Table 4.3: C ivS0 Troughs Quasar C iv v v ∆v v [1] v [2] d [1] σ [1] S0 max min cent cent BAL dBAL ··· ID TroughID (kms−1) (kms−1) (kms−1) (kms−1) (kms−1) Q3 CS0-1 10477.0 5841.4 4635.7 8427.4 8285.0 0.491 0.032 Q4 CS0-2 −12444.7 −4858.5 7586.2 −9558.5 −9262.8 0.697 0.025 Q10 CS0-3 −15491.2 −3000.0 12491.2 −10869.6 −10629.1 0.531 0.024 Q11 CS0-4 −11596.2 −5638.0 5958.2 − 8516.9 − 8806.4 0.217 0.010 Q12 CS0-5 −10921.6 −3000.0 7921.6 −8088.0 −8031.0 0.648 0.023 − − − − ∆EW ∆EW dBAL[2] σdBAL [2] EW[1] σEW[1] EW[2] σEW[2] ∆EW σ∆EW σ hEWi hEWi (A)˚ (A)˚ (A)˚ (A)˚ (A)˚ (A)˚ 0.442 0.029 11.51 0.56 10.29 0.15 1.22 0.58 0.11 0.06 0.744 0.023 26.86 0.20 28.43 0.10− 1.56 0.22− 0.06 0.01 0.543 0.021 31.37 0.92 35.28 0.73 3.91 1.18 0.12 0.05 0.253 0.016 6.63 0.15 7.73 0.19 1.10 0.24 0.15 0.05 0.593 0.024 27.95 0.32 25.64 0.22 2.31 0.39 0.09 0.02 − − Si iv v v ∆v v [1] v [2] d [1] σ [1] S0 max min cent cent BAL dBAL ··· Trough ID (kms−1) (kms−1) (kms−1) (kms−1) (kms−1) SS0-1 12065.5 5872.0 6193.5 8972.6 8908.6 0.243 0.015 SS0-2 −12525.3 −4926.5 7598.8 −8905.7 −8832.4 0.314 0.016 SS0-3 −12457.4 −3927.3 8530.1 −8587.1 −8601.9 0.400 0.020 SS0-4 −11062.2 −5638.0 5424.2 −8258.4 −8293.8 0.206 0.009 SS0-5 − 5804.7 −3000.0 2804.7 −4427.8 −4462.2 0.503 0.029 − − − − ∆EW ∆EW dBAL[2] σdBAL [2] EW[1] σEW[1] EW[2] σEW[2] ∆EW σ∆EW σ hEWi hEWi (A)˚ (A)˚ (A)˚ (A)˚ (A)˚ (A)˚ 0.137 0.006 6.41 0.68 3.91 0.19 2.50 0.71 0.48 0.14 0.361 0.018 10.84 0.24 12.61 0.13− 1.76 0.27− 0.15 0.03 0.515 0.018 15.14 0.64 20.22 0.48 5.08 0.80 0.29 0.06 0.186 0.010 5.22 0.14 4.70 0.17 0.52 0.22 0.11 0.06 0.467 0.031 6.78 0.20 6.25 0.11 −0.52 0.23 −0.08 0.05 − − Notes. Throughout this table [1] indicates the first-epoch spectra and [2] indicates the second-epoch spectra. (This table is available in its entirety in a machine-readable form in the online journal. A portion is shown here for guidance regarding its form and content.) 124 iv Table 4.4: C SA Troughs Quasar C iv v v ∆v v [1] v [2] d [1] σ [1] SA max min cent cent BAL dBAL ··· ID TroughID (kms−1) (kms−1) (kms−1) (kms−1) (kms−1) Q1 CSA-1 17205.7 3487.4 13718.4 10850.0 10770.0 0.521 0.018 Q5 CSA-2 −20000.0 −4805.6 15194.4 −13846.9 −13939.7 0.755 0.011 Q8 CSA-3 −20000.0 −3000.0 17000.0 −11866.5 −11868.8 0.450 0.015 Q20 CSA-4 −20000.0 −6837.8 13162.2 −13057.2 −12687.1 0.607 0.014 Q28 CSA-5 −20000.0 −7283.0 12717.0 −14376.7 −14356.4 0.291 0.014 − − − − ∆EW ∆EW dBAL[2] σdBAL [2] EW[1] σEW[1] EW[2] σEW[2] ∆EW σ∆EW σ hEWi hEWi (A)˚ (A)˚ (A)˚ (A)˚ (A)˚ (A)˚ 0.527 0.018 33.47 0.80 36.51 0.46 3.04 0.93 0.09 0.04 0.752 0.011 58.06 0.84 57.61 0.45 0.45 0.95 0.01 0.02 0.532 0.014 38.15 1.08 45.28 0.57− 7.13 1.22− 0.17 0.04 0.634 0.013 39.57 0.78 41.52 0.73 1.95 1.07 0.05 0.04 0.306 0.015 19.46 0.42 20.19 0.22 0.73 0.47 0.04 0.03 Si iv v v ∆v v [1] v [2] d [1] σ [1] SA max min cent cent BAL dBAL ··· Trough ID (kms−1) (kms−1) (kms−1) (kms−1) (kms−1) SSA-1 13895.2 4807.7 9087.5 8736.4 8724.1 0.434 0.021 SSA-2 −20000.0 −5385.4 14614.6 −14178.8 −14518.4 0.617 0.017 SSA-3 −15485.3 −3361.3 12124.0 − 9281.0 − 8879.0 0.432 0.016 SSA-4 −19792.4 −6932.7 12859.7 −13757.4 −13620.0 0.426 0.012 SSA-5 −13809.3 −7290.5 6518.8 −10512.4 −10490.5 0.357 0.018 − − − − ∆EW ∆EW dBAL[2] σdBAL [2] EW[1] σEW[1] EW[2] σEW[2] ∆EW σ∆EW σ hEWi hEWi (A)˚ (A)˚ (A)˚ (A)˚ (A)˚ (A)˚ 0.423 0.018 15.58 0.77 17.40 0.37 1.82 0.85 0.11 0.07 0.521 0.020 41.32 0.92 31.13 0.71 10.19 1.16 0.28 0.04 0.524 0.017 21.58 0.91 28.62 0.43− 7.05 1.01− 0.28 0.06 0.467 0.010 24.58 1.03 27.40 0.82 2.83 1.32 0.11 0.07 0.415 0.018 10.61 0.24 12.43 0.11 1.83 0.26 0.16 0.03 125 Table 4.4–Continued Al iii v v ∆v v [1] v [2] d [1] σ [1] SA max min cent cent BAL dBAL ··· Trough ID (kms−1) (kms−1) (kms−1) (kms−1) (kms−1) ASA-1 13232.8 8928.5 4304.3 11126.1 11012.0 0.205 0.016 ASA-2 −10920.5 −7369.8 3550.7 − 9144.0 − 9100.4 0.124 0.010 ASA-3 −12641.1 −9903.0 2738.1 −11232.1 −11231.1 0.148 0.012 ASA-4 −16454.8 −7915.9 8538.9 −12067.3 −12139.7 0.215 0.011 ASA-5 −12102.2 −9794.0 2308.2 −10935.6 −10939.9 0.136 0.009 − − − − ∆EW ∆EW dBAL[2] σdBAL [2] EW[1] σEW[1] EW[2] σEW[2] ∆EW σ∆EW σ hEWi hEWi (A)˚ (A)˚ (A)˚ (A)˚ (A)˚ (A)˚ 0.181 0.009 5.10 0.40 4.75 0.31 0.35 0.51 0.07 0.14 0.136 0.008 2.17 0.49 3.00 0.31− 0.83 0.58− 0.32 0.32 0.164 0.011 2.31 0.35 2.78 0.26 0.47 0.43 0.18 0.24 0.267 0.007 10.12 1.11 13.65 0.96 3.53 1.47 0.30 0.18 0.167 0.007 1.95 0.11 2.38 0.07 0.43 0.14 0.20 0.09 Notes. Throughout this table [1] indicates the first-epoch spectra and [2] indicates the second-epoch spectra. (This table is available in its entirety in a machine-readable form in the online journal. A portion is shown here for guidance regarding its form and content.) 126 150 100 C IV (852) C IV (113) 00 C IV (246) S0 C IV (95) 50 SA Number of BAL Troughs 0 1 1.5 2 2.5 3 3.5 4 4.5 Rest−Frame ∆t (yr) Fig. 4.4: The rest-frame ∆t distributions for all C iv troughs in our main sample (solid iv iv iv black) and C 00 (dot-dashed blue), C S0 (dashed green), and C SA (solid red) BAL troughs. The total number of BAL troughs in each sample is given in parentheses. After renormalization for the number of each type of trough, there are no statistically significant differences between the distributions. Given their definitions, the sum of the total numbers iv iv iv iv of C 00,C S0, and C SA troughs are not necessarily equal to total number of C troughs in our main sample. The Anderson-Darling test results indicate no significant ∆t differences between the samples. 127 which is the mean velocity where each data point is weighted with its distance from the normalized continuum level. We adapt redshift values from Hewett & Wild (2010) for all velocity calculations. 4.3.3 Comparisons with Mg ii, Fe ii, and P v We have compared our adopted classification of C iv BAL troughs with the standard subtypes of HiBALs, LoBALs, and FeLoBALs (see Section 4.1). All 113 C iv00 and 246 C ivS0 troughs fit into the definition of HiBALs where only BAL troughs of high-ionization transitions are present in the spectra. Standard classes of LoBALs and FeLoBALs are iden- tified with the presence of Mg ii and Fe ii absorption, respectively. We visually investigated ii ii iv the Mg λλ2797, 2804 A˚ and Fe λλ2400, 2600 A˚ absorption-line regions for 95 C SA troughs to examine the correspondence to the standard LoBAL and FeLoBAL definitions. iv ii The spectra of 34 quasars with C SA troughs do not have coverage of the Mg region, and these are not utilized in our correspondence checking. We find that 74% (45 out of iv ii ≈ 61) of quasars with C SA troughs exhibit Mg BAL/mini-BAL troughs at corresponding velocities; four of them also exhibit Fe ii BAL/mini-BAL troughs. There are 16 quasars with C ivSA troughs that have no Mg ii absorption. Due to the differences between their ionization potentials (28.4 eV for Al iii and 15.0 eV for Mg ii), Al iii absorption is expected to be slightly more common than Mg ii absorption given that higher ionization troughs are found more frequently than lower ionization troughs in BAL quasars (e.g., Hall et al. 2002). Previous investigations of P v λλ1118, 1128 A˚ absorption lines corresponding in velocity with C iv and Si iv BAL troughs have shown that P v absorption lines are an important indicator of line saturation (e.g., Hamann 1998; Arav et al. 2001); note that the ionization potentials of P v (65.0 eV) and C iv (64.5 eV) are very similar. We thus visually v iv iv iv investigate the P BAL regions that align with our C 00,C S0, and C SA troughs. The spectral coverage is sufficient to investigate the P v BAL region for 40 C iv00 troughs, 113 C ivS0 troughs, and 47 C ivSA troughs. Our visual inspection reveals that spectra of a large fraction ( 88%) of C iv troughs exhibit visually detectable P v absorption ≈ SA features; a majority ( 70%) of C iv troughs are accompanied by moderate-to-strong P v ≈ SA absorption. P v absorption that aligns with C ivS0 troughs is generally weaker than that aligning with C ivSA troughs. Approximately half of the C ivS0 troughs are accompanied by detectable P v absorption; however, only 10% of those C iv troughs align with ≈ S0 moderate-to-strong P v absorption. Only a small fraction ( 12%) of C iv troughs are ≈ 00 accompanied by detectable P v absorption, and the strength of this P v absorption is generally weak. These results show that quasars possessing detectable P v absorption in their spectra are also more likely to present lower-ionization transitions; therefore their C iv troughs are likely to be C ivSA or to a smaller extent C ivS0 troughs. 128 4.4 Results In this section, we present the observational results of our investigation. Utilizing iv iv iv the two-epoch observations for 113 C 00, 246 C S0, and 95 C SA BAL troughs, we investigate the C iv, Si iv, and Al iii BAL profiles (Section 4.4.1), the C iv trough strengths (Section 4.4.2), the C iv trough velocities (Section 4.4.3), the C iv trough variation profiles (Section 4.4.4), the C iv trough EW variation characteristics (Section 4.4.5), and the C iv, Si iv, and Al iii trough EW variation correlations (Section 4.4.6). 4.4.1 BAL-Trough Profiles It is well known that BAL troughs from different transitions show different profiles; for instance, Al iii BAL troughs tend to be narrower and align with lower velocity portions of corresponding C iv BAL troughs (e.g., Weymann et al. 1991; Voit et al. 1993; Trump et al. 2006). Moreover, previous studies (e.g., Weymann et al. 1991; Reichard et al. 2003; Allen et al. 2011) have demonstrated that C iv BAL troughs tend to be stronger in quasars exhibiting Al iii and/or Mg ii BAL troughs (i.e., LoBAL quasars). To compare the typical properties of C iv00,C ivS0, and C ivSA BAL troughs, we calculate composite mean profile shapes for each C iv group. Figures 4.5 and 4.6 show the mean profile shapes for 113 C iv00, 246 C ivS0, and 95 C ivSA BAL troughs as a function of outflow velocity relative to vmin and as a function of normalized fractional trough width, respectively. We calculate the mean profiles by setting the outflow velocity of each BAL trough to 0 at its vmin. We iv fix the normalized flux density, Fλ, to 1 at velocities higher than vmax for each C BAL trough. We calculate normalized C iv trough width by fixing the width of each C iv trough to 100 at its maximum blue-edge velocity and to 0 at its minimum red-edge velocity. We also present mean profile shapes of Si iv and Al iii BAL troughs in overlapping velocity ranges for comparison. iv A comparison of the resulting composite profiles indicates that C 00 troughs tend to be shallower and narrower than both C ivS0 and C ivSA troughs. The C ivS0 and C ivSA trough depths change in a characteristic manner as a function of velocity within the trough: troughs are generally deeper at lower velocities. Consistent with previous studies, the composite profiles show that C iv BALs are usually deepest when there is an Al iii BAL at corresponding velocities. Similarly, a Si iv BAL-trough profile change is apparent between the C ivS0 and C ivSA samples. We also find that Al iii BAL troughs tend to be narrower iv and be found at lower velocities than corresponding C SA BAL troughs. We find qualitatively consistent results when computing median composites instead of mean composites. This indicates that outliers are not strongly affecting our composites. 4.4.2 C iv BAL-Trough Strengths iv iv iv For a more quantitative comparison between the C 00,C S0, and C SA groups, we assess differences between measured C iv BAL-trough properties in this subsection and 129 1 0.8 C IV 0.6 0.4 C IV 00 ) 0.2 λ 1 Normalized C IV BAL width (%) 0.8 0.6 Si IV 0.4 C IV C IV 0.2 S0 Normalized Flux Density (F 1 0.8 Si IV Al III 0.6 0.4 C IV C IV 0.2 SA 14000 12000 10000 8000 6000 4000 2000 0 ∆v (km s−1) Fig. 4.5: Composite BAL-trough profiles for C iv00 (top), C ivS0 (middle), and C ivSA (bottom) BAL troughs. Solid curves show mean normalized flux density for C iv (blue), Si iv (green), and Al iii (red) BAL troughs at overlapping velocities as a function of out- flow velocity relative to vmin. The horizontal dashed black lines in each panel show the level for 10% under the continuum, a level important for BAL-trough identification (See iv iv iv Equation 4.1). The composite profile shapes of C 00,C S0, and C SA troughs differ significantly; stronger and wider C iv troughs are found when BAL troughs from lower ionization transitions are present. 130 1 0.8 C IV 0.6 0.4 C IV 00 ) 0.2 λ 1 Normalized C IV BAL width (%) 0.8 Si IV 0.6 C IV 0.4 C IV 0.2 S0 Normalized Flux Density (F 1 0.8 Si IV Al III 0.6 C IV 0.4 C IV 0.2 SA 100 90 80 70 60 50 40 30 20 10 0 Normalized C IV BAL width (%) Fig. 4.6: Same as Figure 4.5 but the x-axes show normalized C iv trough width as a percentage; 0 corresponds to the smallest blueshift velocity and 100 to the largest blueshift iv iv velocity of a given trough. The C S0 and C SA trough depth changes as a function of velocity within the trough: troughs are generally deeper at lower velocities. 131 1 60 C IV 00 C IV S0 0.8 50 C IV SA 40 0.6 30 0.4 20 Number of BAL Troughs Fraction of BAL Troughs 0.2 10 0 0 0 20 40 60 0 20 40 60 ˚ Fig. 4.7: Average EW, EW , distributions for C iv (dot-dashed blue), C iv (dashed h i 00 S0 green), and C ivSA (solid red) BAL troughs. The right panel shows the fraction of C iv00, C ivS0, and C ivSA BAL troughs to all 852 C iv BAL troughs in our sample. Since troughs with mini-BALs are excluded, the numbers in the right panel need not sum to unity for a given bin. The EW distributions of the three C iv groups are significantly (99.9%) h i different from each other. iv iv the next. Figure 4.7 shows the distributions of average EW, EW , for C 00,C S0, iv h i and C SA BAL troughs. Considering that the total number of BAL troughs increases with decreasing EW , the fraction of BAL troughs with given EW is also displayed in h i h i iv iv Figure 4.7. We calculate the fractions as the ratio of the number of C 00,C S0, and iv iv C SA BAL troughs to all 852 main-sample C troughs; they therefore need not sum to unity in a given bin. The mean EW is 15.46 0.42 A˚ for all 852 C iv BAL troughs in our main sample, h i ± whereas it is 4.76 0.25 A˚ for C iv , 19.29 0.62 A˚ for C iv , and 32.38 1.35 A˚ for C iv ± 00 ± S0 ± SA troughs as given in Table 4.5. Uncertainties on the mean are calculated using the standard σ/√N formula. The median EW is 11.51 A˚ for all C iv, 4.44 A˚ for C iv , 17.58 A˚ h i 00 for C ivS0, and 31.14 A˚ for C ivSA BAL troughs. These results confirm and quantify the increase of C iv BAL-trough strength with the existence of absorption lines from lower ionization-level transitions. In order to determine the contributions of the depth and width components of BAL EWs, we assess average depth from two-epoch observations, dBAL , and velocity width, iv iv iv h i ∆v, distributions for C 00,C S0, and C SA troughs (see Figure 4.8). Table 4.5 presents the mean d and ∆v values for all three C iv BAL-trough groups. We compare these h BALi 132 Table 4.5: Average values of BAL-trough properties for C iv00,C ivS0,C ivSA, and all C iv BAL troughs iv iv iv iv C 00 C S0 C SA All C EW (A)˚ 4.76 0.25 19.29 0.62 32.38 1.35 15.46 0.42 h i ± ± ± ± d 0.25 0.009 0.47 0.009 0.56 0.012 0.41 0.006 h BALi ± ± ± ± ∆v (km s−1) 3940 182 8600 255 11668 428 7195 146 ± ± ± ± Number of Data Points 113 246 95 852 Uncertainties on the mean are calculated using the standard σ/√N formula. distributions using an AD test and find that both the dBAL and ∆v distributions for iv iv iv h i C 00,C S0, and C SA BAL troughs are significantly different (at a confidence level iv of > 99.9%) from each other. Our findings indicate that C SA BAL troughs tend to be iv the deepest and widest BAL troughs, while C 00 troughs tend to be the shallowest and narrowest. The ranges of the d and ∆v values for C iv ,C iv , and C iv BAL h BALi 00 S0 SA troughs demonstrate that the contributions of the depth and the width to the differences between BAL-trough EWs are comparable; the d values change by a factor of 2.3 h BALi ≈ and the ∆v values change by a factor of 3 between the C iv and C iv samples. ≈ 00 SA As can be seen in Figures 4.5 and 4.6, the strength of Si iv BAL troughs is also larger when Al iii BAL troughs are present. The mean EW is 7.35 0.49 A˚ for Si iv h i ± BAL troughs in the C iv sample and 20.84 1.02 A˚ for Si iv BAL troughs in the C iv S0 ± SA sample. The mean EW values change by a factor of 2.8, which is even stronger than h i ≈ the corresponding change for C iv of 1.7. ≈ 4.4.3 C iv BAL-Trough Velocities To assess differences in BAL-trough velocities, we investigate minimum velocity, v , maximum velocity, v , and average centroid velocity, v , distributions for min max h centi C iv ,C iv , and C iv BAL troughs (see Figure 4.9). The mean v , v , and v 00 S0 SA min max h centi values are given in Table 4.6. AD test results show that all three vmin distributions are significantly different (at a level of > 99.9%) from each other. C iv00 BAL troughs tend to have higher onset velocities than C iv and C iv troughs. The mean v values change by a factor of 2.5 from S0 SA min ≈ C iv00 to C ivSA troughs. Such differences between the vmin distributions for C iv00,C ivS0, 133 1 C IV 00 60 C IV S0 0.8 C IV 50 SA 40 0.6 30 0.4 20 Number of BAL Troughs Fraction of BAL Troughs 0.2 10 0 0 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 1 45 40 0.8 35 30 0.6 25 20 0.4 15 Number of BAL Troughs Fraction of BAL Troughs 10 0.2 5 0 0 5000 10000 15000 5000 10000 15000 −1 ∆v (km s ) ∆v (km s−1) Fig. 4.8: Average depth, d (upper panels), and velocity width, ∆v (lower panels), h BALi distributions for C iv00 (dot-dashed blue), C ivS0 (dashed green), and C ivSA (solid red) BAL troughs. The right panels show the fraction of C iv00,C ivS0, andC ivSA BAL troughs relative to all C iv BAL troughs in our main sample. The d and ∆v distributions for h BALi C iv00,C ivS0, and C ivSA BAL troughs are significantly (99.9%) different from each other. 134 1 80 C IV 00 C IV 70 S0 0.8 C IV 60 SA 50 0.6 40 0.4 30 Number of BAL Troughs 20 Fraction of BAL Troughs 0.2 10 0 0 −20000 −15000 −10000 −5000 −20000 −15000 −10000 −5000 −1 v (km s ) v (km s−1) min min 1 120 100 0.8 80 0.6 60 0.4 40 Number of BAL Troughs Fraction of BAL Troughs 0.2 20 0 0 −20000 −15000 −10000 −5000 −20000 −15000 −10000 −5000 −1 v (km s ) v (km s−1) max max 1 60 0.8 50 40 0.6 30 0.4 20 Number of BAL Troughs Fraction of BAL Troughs 0.2 10 0 0 −20000 −15000 −10000 −5000 −20000 −15000 −10000 −5000 Fig. 4.9: Minimum velocity, vmin (upper panels), maximum velocity, vmax (middle panels), and average centroid velocity, v (lower panels), distributions for C iv (dot-dashed h centi 00 blue), C ivS0 (dashed green), and C ivSA (solid red) BAL troughs. The right panels display iv iv iv iv the fraction of C 00,C S0, and C SA BAL troughs relative to all C BAL troughs in our main sample. The v and v distributions for C iv ,C iv , and C iv BAL min h centi 00 S0 SA troughs are significantly (99.9%) different from each other, whereas the vmax distributions do not show highly significant differences. 135 Table 4.6: Average values of BAL-trough velocities for C iv00,C ivS0,C ivSA, and all C iv BAL troughs iv iv iv iv C 00 C S0 C SA All C v (km s−1) 11898 400 7434 238 4838 241 8424 154 min − ± − ± − ± − ± v (km s−1) 15836 426 16035 308 16506 453 15619 168 max − ± − ± − ± − ± v (km s−1) 13830 402 11998 235 11405 291 12229 140 h centi − ± − ± − ± − ± Number of Data Points 113 246 95 852 Uncertainties on the mean are calculated using the standard σ/√N formula. iv and C SA BAL troughs are perhaps expected given that the average outflow velocity is higher for weak BAL troughs (e.g., Weymann et al. 1991). Unlike the vmin distributions, the vmax distributions notably do not show highly iv iv significant differences. AD test results show that P is 96.4% for the C 00 vs. C S0 samples, 37.9% for the C ivS0 vs. C ivSA samples, and 91.8% for the C iv00 vs. C ivSA samples. This result apparently arises due to the significant compensatory differences in BAL-trough widths and onset velocities; wide BAL troughs tend to have low onset velocities and narrow BAL troughs tend to have high onset velocities. AD test results show that the v distributions are significantly different (at a h centi level of > 99.9%) from each other. C ivSA BAL troughs tend to be found at the lowest such iv velocities, and C 00 troughs tend to be found at the highest such velocities, although the mean v values differ by only 25%. h centi ≈ 4.4.4 C iv BAL-Variation Profiles We assess the overall differences between C iv BAL-variation characteristics by com- paring composite variation profiles for C iv00,C ivS0, and C ivSA troughs. First, we calcu- late the absolute value of the depth variation between the SDSS and BOSS spectra for each data point of each trough. Second, we construct the mean composites by fixing each C iv trough width to 100 at its maximum blue-edge velocity and to 0 at its minimum red-edge velocity. Figure 4.10 displays the mean composites of depth-variation profiles for C iv00, C ivS0, and C ivSA BAL troughs. We find qualitatively consistent results when computing median composites instead of mean composites. Figure 4.10 also presents the mean com- posite for absolute fractional variations which is the variation of depth between SDSS and BOSS divided by the average depth. 136 0.6 C IV 0.1 00 0.4 0.05 C IV 0.2 00 0 0 0.6 C IV 0.1 S0 0.4 0.05 C IV 0.2 S0 0 0 C IV 0.6 SA Mean BAL−Trough Depth Variation 0.1 0.4 Mean Fractional BAL−Trough Depth Variation 0.05 C IV 0.2 SA 0 0 100 80 60 40 20 0 100 80 60 40 20 0 Normalized C IV BAL width (%) Normalized C IV BAL width (%) Fig. 4.10: BAL-trough variation profiles for C iv00 (top panel), C ivS0 (middle panel), and iv C SA (bottom panel). Left panels: Solid-black curves show the mean composite profiles for absolute depth variations of C iv BAL troughs as a function of normalized BAL-trough width. The light gray curves show the error on the mean. Right panels: Mean composite profile for absolute fractional depth variations of C iv BAL troughs, where the fractional iv variation is the depth variation divided by the average depth. While the C 00 troughs show similar variability across the entire trough, the C ivS0 and C ivSA troughs show less variability at lower velocities where the Si iv and Al iii absorption tend to be the strongest. 137 iv The composite variation profiles demonstrate that the level of variability for C 00 troughs is roughly uniform across the BAL-trough width. Although the variation profile iv iv for C S0 troughs is not strongly different from that of C 00 troughs, the level of variation iv generally tends to be smaller for the lower velocity portions of C S0 troughs (up to 30% of the normalized width). The lower velocity portions of C iv troughs also usually ≈ S0 correspond to regions where the Si iv BAL troughs are strongest, as shown in Figures 4.5 and 4.6. Figure 4.10 shows that the variation profile for C ivSA troughs is strongly different from the variation profiles for C iv00 and C ivS0 troughs. The lower velocity portions (up to 80% of the normalized width) of C iv troughs show notably less variability than ≈ SA matched portions of C iv00 and C ivS0 troughs; this result is highly statistically significant, considering the number of independent spectral data points and their error bars. Figures 4.5 iv and 4.6 demonstrate that the lower velocity portions of C SA troughs correspond to regions where Al iii BAL troughs are generally found. The higher velocity portions of iv iv iv C SA troughs are more consistent with matched portions of C 00 and C S0 troughs. Composite fractional-variation profiles further highlight the differences between the variation characteristics of C iv00,C ivS0, and C ivSA BAL troughs. These arguably allow the best possible comparison of overall variability between these C iv groups, since they account for the significantly different strengths of C iv00,C ivS0, and C ivSA troughs (see Section 4.4.2). Overall, we find that C iv00 troughs tend to be significantly more fractionally variable than both C ivS0 and C ivSA troughs, and likewise C ivS0 troughs tend to be somewhat more fractionally variable than C ivSA troughs. Variability is again smallest in iv iv the lower velocity portions of C S0 and C SA troughs. 4.4.5 C iv BAL EW Variability 4.4.5.1 EW Variability as a Function of EW h i In this section, we investigate the characteristics of BAL EW variability assessing iv iv iv the differences between C 00,C S0, and C SA troughs. Figure 4.11 presents the EW variation, ∆EW, between the two-epoch spectra as a function of average EW, EW . h i For comparison, the standard-deviation curves for these three C iv groups are also given; the standard-deviation curves are calculated using a sliding window where each window contains 10% of the total number of BAL troughs found in a given group (i.e., 11 for C iv00, 25 for C ivS0, and 10 for C ivSA). We statistically remove the mean EW error in each window from the standard deviation, so the dispersion shown in Figure 4.11 refers to the intrinsic dispersion. Consistent with Section 4.5 of Filiz Ak et al. (2013), we find that the standard de- viation of ∆EW generally increases with increasing EW for all C iv groups. Given that h i such a trend exists for all C iv groups, a proper intergroup comparison requires considera- tion of EW variation for troughs with similar EW values. The C iv ∆EW spread is less h i S0 than that for C iv BALs with similar EW values (i.e., 7 < EW < 10 A).˚ Similarly, 00 h i h i the C iv ∆EW spread is less than that for C iv BALs for 12 < EW < 35 A.˚ These SA S0 h i 138 results suggest that C iv EW variation has a dependence upon the existence of absorption lines from Si iv and Al iii transitions. In Figure 4.12, we show the fractional EW variation, ∆EW/ EW , as a function h i of EW for C iv ,C iv , and C iv troughs. Similarly to Figure 4.11, solid curves h i 00 S0 SA represent the sliding-window standard deviations. The curves indicate that the spread of ∆EW/ EW for C iv is larger than that for C iv , and the ∆EW/ EW spread for h i 00 S0 h i C iv is larger than that for C iv in matched EW ranges. S0 SA h i Previous BAL-variability studies (Gibson et al. 2008, 2010; Capellupo et al. 2011; Filiz Ak et al. 2013) have demonstrated that BAL EW variability increases with increasing timescale. However, the trends in Figures 4.11 and 4.12 are not caused or amplified by this iv iv effect; the testing in Section 4.3 showed that the timescale distributions for C 00,C S0, iv and C SA troughs do not have any significant differences from each other (see Figure 4.4). Considering that the EW distributions are significantly different between the three h i C iv groups (see Figure 4.7) and that there are strong correlations between ∆EW or ∆EW/ EW and EW for all C iv troughs (also see Section 4.5 of Filiz Ak et al. 2013), | h i| h i we investigate the EW variation distributions of these three groups using matched samples of troughs with similar EWs. 4.4.5.2 BAL-Trough Samples with Matching EWs We investigated the EW-variation characteristics of C iv00 vs.C ivS0 and C ivS0 vs.C ivSA troughs using matched samples. These matched samples contain the same num- ber of BAL troughs from both groups matched by first-epoch EW, EW1. Requiring similar iv EW1 for each trough pair allows comparison of variation behaviors for C groups having iv iv similar initial conditions. We first randomly select a C 00 trough for each C S0 trough such that their EWs are in agreement to within 1σ. Due to significant differences in the EW distributions, we cannot match all data points of a group. In the matching procedure, we sample any given BAL trough only once. Second, we applied a two-sample AD test to compare the ∆EW distributions. We repeated the matching and AD testing 1000 times and found the median results given in Table 4.7. Comparing the ∆EW distributions of C iv00 and C ivS0 BAL troughs with matched EWs (where each distribution has 40 troughs), we found that the two distributions signif- icantly (P > 99.9%) differ from each other. Similarly, we repeat the matching for C ivS0 iv and C SA troughs. The test results for the comparison between the ∆EW distributions iv iv of C S0 and C SA troughs with similar EWs (where each has 65 troughs) show that the two distributions differ with a significance of 97.8%. Figure 4.13 presents the ∆EW iv iv distributions for C 00 and C S0 BAL troughs with matched EWs in the left panel, and iv iv C S0 and C SA BAL troughs with matched EWs in the right panel. As given in Table 4.7, the mean of ∆EW for C iv troughs is 1.3 times larger | | 00 ≈ than that for matched C iv troughs. Likewise, the mean of ∆EW for C iv troughs is S0 | | S0 1.8 times larger than that for matched C iv troughs. The standard deviation of the ≈ SA 139 C IV 15 00 CIV S0 10 CIV SA 5 ˚ A) ( 0 EW ∆ −5 −10 −15 1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 Fig. 4.11: ∆EW vs. EW for C iv (blue circles), C iv (green squares), and C iv (red h i 00 S0 SA triangles) BAL troughs. Solid curves show rms values calculated with a sliding window containing 10% of the total data points. The spread of ∆EW generally increases with iv iv iv increasing EW for C 00,C S0, and C SA troughs. In overlapping EW ranges, iv h i iv h iv i C 00 troughs tend to be more variable than C S0 troughs, and similarly C S0 troughs tend to be more variable than C ivSA troughs. 140 1 0.5 0 EW / 1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 ˚ Fig. 4.12: Same as Figure 4.11 but for ∆EW/ EW . The spread of ∆EW/ EW decreases h i iv h i with increasing EW . In overlapping EW ranges, C 00 troughs tend to be fractionally h i iv h i iv more variable than C S0 troughs, and similarly C S0 troughs tend to be fractionally iv more variable than C SA troughs. ∆EW distribution for C iv troughs is 1.1 times larger than that for matched C iv 00 ≈ S0 troughs, and for C iv troughs is 1.7 times larger than that for matched C iv troughs. S0 ≈ SA Although the matched C ivS0 and C ivSA BAL troughs have both a larger sample size (65) and difference between the σ values (5.16 3.09 = 2.07) than the matched ∆EW − C iv00 and C ivS0 troughs, the AD-test results indicate a more significant overall difference for the ∆EW distributions of C iv00 and C ivS0 troughs. This result is mainly due to a iv iv more significant difference between the mean ∆EW values of matching C 00 and C S0 iv iv troughs (3.6σ) than that for matching C S0 and C SA troughs (1.9σ). Another notable iv result of these comparisons is that the means of the ∆EW distributions for both the C 00 and C iv troughs with similar EWs differ from zero at more than 2σ: 1.13 0.47 A˚ S0 − ± and 1.33 0.49 A,˚ respectively. Moreover, the means of the ∆EW distributions for C iv ± S0 and C ivSA troughs with similar EWs also differ from zero at more than 1σ. The means of the ∆EW distributions for C ivS0 in the two different matching samples change from a positive value to a negative value. These results are initially surprising given the fact that the mean of the ∆EW distribution for all 852 C iv BAL troughs in our main sample is consistent with zero ( 0.0055 0.0373 A;˚ also see Section 4.4 of Filiz Ak et al. 2013). − ± Indeed, the weakening and strengthening of BAL troughs are expected to be balanced to maintain an equilibrium population of BAL troughs in quasar spectra. 141 Table 4.7: Two-Sample Anderson-Darling Test Results for BAL-Trough Samples with Matching First-Epoch EWsa iv iv iv iv C 00 C S0 C S0 C SA Mean ∆EW 1.13 0.47 1.33 0.49 0.70 0.64 0.75 0.38 − ± ± − ± ± Mean ∆EW 2.34 0.34 1.84 0.45 3.78 0.44 2.11 0.29 | | ± ± ± ± σ 3.42 0.34 3.14 0.35 5.16 0.45 3.09 0.27 ∆EW ± ± ± ± pAD > 99.9% 97.8% Sample Size 40 65 aEntries are the median values calculated for 1000 random draws. Uncertainties on the mean and standard deviation are calculated using the standard σ/√N formula and Equation 14.1.3 of Press et al. (1992), respectively. The differences between the mean values of the ∆EW distributions can occur, for instance, if not only ∆EW but also ∆EW has a dependence on BAL-trough strength. To | | assess such dependency, we show ∆EW as a function of first-epoch EW for C iv00,C ivS0, and C ivSA troughs in Figure 4.14. The best fits of basic linear-regression models are plotted in each panel of Figure 4.14 to demonstrate the apparent trends. Using Spearman rank-correlation tests (see Press et al. 1992), we find connections between ∆EW and EW1 for C iv00 (P > 99.9%) and likely also for C ivS0 (P = 98.4%) troughs. The apparent ∆EW vs. EW1 relation primarily arises because the BAL-trough EW variation range is limited. Given that a BAL trough cannot weaken by more than its first- epoch EW and cannot strengthen by more than its second-epoch EW, EW2, the variation strength is restricted to be EW2 ∆EW EW1. In addition, we exclude quasars ≥ ≥ − 20 that do not satisfy our necessary quasar-selection criterion of BI3 > 0 in either epoch (see Section 4.2). Thus, this criterion leads to exclusion of disappearing (i.e., ∆EW EW ; ≈− 1 e.g., Filiz Ak et al. 2012) and emerging (i.e., ∆EW EW ) BAL troughs from our main ≈ 2 sample. In our matched samples, we must compare the ∆EW distribution of the stronger C iv00 troughs with that of the weaker C ivS0 troughs, and that of the stronger C ivS0 troughs with that of the weaker C ivSA troughs. Therefore the differences in the means of these ∆EW distributions are expected given the relations shown in Figure 4.14. 4.4.6 EW Variation Correlations Previous studies (e.g., Gibson et al. 2010; Capellupo et al. 2012; Filiz Ak et al. 2013) have demonstrated that Si iv BAL troughs tend to vary in concert with C iv troughs at corresponding velocities and that the strength of fractional EW variations tends to be larger 142 20 20 C IV C IV S0 00 15 C IV 15 C IV SA S0 10 10 5 5 Number of BAL Troughs Number of BAL Troughs 0 0 −10 −5 0 5 10 −10 0 10 ∆EW( A)˚ ∆EW( A)˚ iv iv iv Fig. 4.13: ∆EW distributions for C 00 vs. C S0 troughs (left panel) and for C S0 iv vs. C SA troughs (right panel), where the samples show BAL troughs with similar EWs. iv iv iv C 00 troughs are more variable than C S0 troughs, while C S0 troughs are in turn more iv variable than C SA troughs. for Si iv BAL troughs. In order to investigate similar relations for C ivS0 (Section 4.4.6.1) and C ivSA (Section 4.4.6.2) troughs, we assess variation correlations between C iv troughs and corresponding Si iv and Al iii troughs. 4.4.6.1 EW Variation Correlations for the C ivS0 Sample Figure 4.15 presents the ∆EW and ∆EW/ EW correlations for C iv and the h i S0 corresponding Si iv troughs. Consistent with the findings of Filiz Ak et al. (2013) for all C iv troughs, the Spearman-test results show that both the ∆EW and ∆EW/ EW h i correlations are highly significant (P > 99.9%). Given the apparent scatter in both panels of Figure 4.15, we determine the best fits using a Bayesian linear-regression model that considers the intrinsic scatter of the sample (Kelly 2007). We found the following relations, calculating the mean and the standard deviation of the linear-regression model parameters for 10,000 random draws from the sample: ∆EW = (1.20 0.082) ∆EW + (0.051 0.216), σ = 3.4 A˚ (4.2) C IVS0 ± × Si IV ± IS ∆EW ∆EW = (0.515 0.023) + (0.006 0.011), σ = 0.15. (4.3) EW ± × EW ± IS h i C IVS0 h i Si IV Here σIS is the standard deviation of the intrinsic scatter. The slopes of the ∆EW and ∆EW/ EW correlations are consistent with the findings in Section 4.6 of Filiz Ak et al. h i 143 15 C IV 00 C IV C IV S0 SA 10 5 A) ˚ ( 0 EW ∆ −5 disappearance −10 −15 0 5 10 15 0 20 40 60 0 20 40 60 EW ˚ EW ( ˚A) EW ( ˚A) 1( A) 1 1 iv iv iv Fig. 4.14: ∆EW as a function of EW1 for C 00 (left), C S0 (middle), and C SA (right) BAL troughs. The solid lines show the best fits of basic linear-regression models in each panel. The dashed lines denote where ∆EW = EW (corresponding to BAL disappear- − 1 ance). The apparent connection between ∆EW and EW1 arises primarily because a BAL trough cannot weaken by more than its first-epoch EW. (2013) within 1σ, indicating that the EW variations of C ivS0 troughs are comparable to those of Si iv troughs, whereas the fractional EW variations of C ivS0 troughs are approx- imately half of those of Si iv troughs. 4.4.6.2 EW Variation Correlations for the C ivSA Sample Figure 4.16 displays the ∆EW and ∆EW/ EW relations for C iv BAL troughs h i SA and the corresponding Si iv and Al iii BAL troughs. The Spearman-test results indicate highly significant (P > 99.9%) correlations for both ∆EW and ∆EW/ EW between C iv h i SA and the corresponding Si iv troughs. The ∆EW correlation between C ivSA and the cor- responding Al iii troughs is marginally significant (P = 99.8%), while the ∆EW/ EW h i correlation is highly significant (P > 99.9%). The ∆EW correlation between C ivSA and Al iii troughs is not highly significant, possibly because of the small range of ∆EW values for both ions. In addition to this effect, the relatively small number of data points in the iv C SA sample, and the apparently large intrinsic scatter of the ∆EW values, may hide any possible correlation. The following relations are found using the Bayesian linear-regression fit: ∆EW = (0.646 0.081) ∆EW + (0.080 0.280), σ = 2.5 A˚ (4.4) C IVSA ± × Si IV ± IS 144 15 C IV C IV S0 1 S0 10 0.5 ˚ A) 5 C IV ( CIV 0 0 EW ∆ −5 EW/ −10 0 10 −1 0 1 ˚ ∆EW ( A) ∆EW/ Fig. 4.15: The ∆EW (left panel) and ∆EW/ EW (right panel) correlations between C iv h i S0 BAL troughs and the corresponding Si iv BAL troughs. Spearman-test results show highly significant correlations (P > 99.9%) for both panels. The solid-blue lines show the best fit found using a Bayesian linear-regression model. ∆EW ∆EW = (0.355 0.044) (0.002 0.008), σ = 0.08 (4.5) EW ± × EW − ± IS h i C IVSA h i Si IV ∆EW = (0.172 0.171) ∆EW + (0.493 0.348), σ = 3.3 A˚ (4.6) C IVSA ± × AlIII ± IS ∆EW ∆EW = (0.111 0.035) + (0.005 0.011), σ = 0.10. (4.7) EW ± × EW ± IS h i C IVSA h i Al III Figure 4.16 presents the best fits of the C iv and Si iv variation relations both for the C ivS0 and the C ivSA samples for comparison purposes. Consistent with the results of our investigations in Section 4.4.5, the ranges of ∆EWC IV and ∆EW/ EW C IV in iv h i iv Figures 4.15 and 4.16 indicate that C SA troughs tend to be less variable than C S0 troughs. Moreover, the ranges of ∆EWSi IV and ∆EW/ EW Si IV in Figures 4.15 and 4.16 iv iv h i suggest that the Si BAL troughs of the C SA sample show less variation than those of iv iv the C S0 sample. The flat slopes of Equations 4.6 and 4.7 indicate that the C SA troughs tend to show very small variations. The slopes of Equations 4.5 and 4.7 indicate that Al iii troughs tend to be more fractionally variable than both the corresponding C ivSA and Si iv 145 C IV C IV SA SA 10 0.4 5 0.2 ˚ A) C IV ( CIV 0 0 EW ∆ EW/ −5 ∆ −0.2 −10 −0.4 −20 −10 0 10 20 −1 −0.5 0 0.5 1 ˚ ∆EW ( A) ∆EW/ C IV C IV SA SA 10 0.4 5 0.2 ˚ A) C IV ( CIV 0 0 EW ∆ EW/ −10 −0.4 −10 −5 0 5 10 −2 −1 0 1 2 ∆EW ˚ ∆EW/ iv Fig. 4.16: The ∆EW (left panels) and ∆EW/ EW (right panels) correlations for C SA iv iv hiii i vs. Si troughs (top panels) and C SA vs. Al troughs (bottom panels). Spearman-test results show highly significant (both with P > 99.9%) ∆EW and ∆EW/ EW correlations h i for C iv vs. Si iv troughs, and a significant ∆EW/ EW correlation for C iv vs. Al iii SA h i SA troughs, whereas the ∆EW correlation for C ivSA vs. Al iii troughs is marginally significant with P = 99.8%. The solid-blue lines show the best-fit relations in each panel. The dashed-black lines in the top two panels show the best-fit models for C ivS0 BAL troughs (Equations 4.2 and 4.3) for comparison. 146 troughs, and Si iv troughs tend to be more fractionally variable than the corresponding iv C SA troughs. 4.5 Summary of Results, Discussion, and Future Work We have investigated the profiles, standard characteristic properties, and variation behaviors of C iv BAL troughs, considering how these change when BAL troughs from Si iv and Al iii are present at corresponding velocities. We have utilized a sample of 852 C iv BAL troughs; 113 of these have no detection of any corresponding Si iv or Al iii BALs or mini-BALs in both epochs (C iv00 troughs), 246 of these are accompanied by a Si iv BAL trough but have no corresponding Al iii BALs or mini-BALs (C ivS0 troughs), and 95 of these are accompanied by both Si iv and Al iii BALs (C ivSA troughs). The main observational findings of our study are the following: iv iv iv 1. The composite profiles of C 00,C S0, and C SA troughs differ significantly; stronger C iv troughs are found when accompanying BAL troughs from lower ion- iv ization transitions are present. Furthermore, the composite profiles for C S0 and C ivSA troughs are deeper at the lower velocities where Al iii and, to a lesser extent, Si iv troughs are preferentially found. See Section 4.4.1. 2. The two-epoch average EW, EW , distributions for C iv00,C ivS0, and C ivSA BAL h i iv troughs are significantly (> 99.9%) different. Generally, C 00 troughs have small iv iv EWs, C S0 troughs have moderate EWs, and C SA troughs have large EWs. We find that increases in both depth and velocity width contribute comparably to the iv iv iv increase in EW from C 00 to C S0 to C SA troughs. See Section 4.4.2. 3. The minimum and average centroid velocities decrease fromC iv00 to C ivS0 to C ivSA troughs; this decrease is most notable for the minimum velocity, which changes on average by a factor of 2.5, while the decrease is mild ( 25%) for the average ≈ ≈ centroid velocity. The maximum velocity does not change substantially, owing to correlation between minimum velocity and trough width. See Section 4.4.3. 4. Composite depth-variation and fractional-depth-variation profiles have been used to investigate the relative variability of C iv00,C ivS0, and C ivSA BAL troughs. BAL variability generally decreases from C iv00 to C ivS0 to C ivSA BALs, particularly in a fractional sense. For C ivS0 and C ivSA troughs, the lower velocity portions of the troughs tend to be the least variable, and these are the regions where Al iii and, to a lesser extent, Si iv troughs are preferentially found. See Section 4.4.4. 5. The spread of ∆EW generally increases with increasing EW , and the spread of h i ∆EW/ EW generally decreases with increasing EW , for C iv ,C iv , and C iv h i h i 00 S0 SA BAL troughs; this result is consistent with the general behavior of all C iv BAL 147 iv troughs (e.g., Filiz Ak et al. 2013). In overlapping ranges of EW ,C 00 troughs iv h iv i appear to vary more strongly than C S0 troughs, and similarly C S0 troughs appear iv to vary more strongly than C SA troughs. See Section 4.4.5.1. 6. For a proper comparison of the variation characteristics of the three C iv groups, we compare ∆EW distributions of samples of BAL troughs with matched first-epoch EWs. C ivS0 troughs are somewhat less variable than C iv00 troughs with matched EWs, and C ivSA troughs are substantially less variable than C ivS0 troughs with matched EWs. See Section 4.4.5.2. iv iv 7. The Si BAL troughs associated with the C S0 sample show EW and fractional EW variations that are generally in concert with those of the corresponding C ivS0 troughs. We quantify the relevant correlations and find that Si iv troughs show similar EW and larger fractional EW variations than corresponding C ivS0 troughs. See Section 4.4.6.1. 8. The EW and fractional EW variations of Al iii troughs are less clearly linked with iv those of C SA troughs, although correlation testing does indicate some correspon- dence. Al iii troughs show larger EW and fractional EW variations than correspond- iv iv ing C SA and Si troughs. See Section 4.4.6.2. We now examine the implications of our observational findings considering the best- developed model for quasar BAL outflows, that of an equatorial radiation-driven disk wind (e.g., Murray et al. 1995; Proga et al. 2000; Higginbottom et al. 2013, see also Section 4.1). While other scenarios for BAL outflows also exist, such as those proposing that BALs primarily are formed at large distances (0.1–10 kpc) from the SMBH (e.g., Arav et al. 2013; also see Faucher-Gigu`ere et al. 2012; but see Section 5.3 of Lucy et al. 2014 for a critique), these have not been developed via numerical simulations to the point where robust comparisons with our observational findings are possible. Of course, any model for BAL winds, current or future, can be usefully constrained by our observational results presented above. Figure 4.17 shows density and poloidal velocity maps of the disk-wind model. In this figure, two lines-of-sight are marked, corresponding to different viewing inclinations, along which an observer would see C iv00 and C ivSA troughs; in parallel with our sub- iv script notation for C troughs, we will refer to these lines-of-sight (LOS) as LOS00 and LOSSA, respectively, throughout. We expect generally correlated changes of ionization level, kinematics, and column density as our line-of-sight is varied from LOS00 to LOSSA (see Section 4.1), and this is relevant to our discussion below.2 We recognize that there 2We appreciate that such correlated changes will have scatter owing to, e.g., time-dependent phenomena leading to local inhomogeneities (see Figure 3 of Proga et al. 2000). Overcoming such scatter is a prime driver for our utilization of large samples in this study. 148 is strong observational evidence supporting intrinsic object-to-object differences as well as inclination effects among BAL quasars (e.g., Boroson & Meyers 1992; Turnshek et al. 1994; Zhang et al. 2010; DiPompeo et al. 2012). For example, the probability of observing iv a C SA trough will be higher for objects having a larger global covering factor of low- ionization gas (cf., the weak [O iii] objects discussed in Boroson & Meyers 1992).3 However, such global-covering-factor effects do not affect our main reasoning below which is focused upon the typical measured properties of C iv00 vs. C ivS0 vs. C ivSA troughs rather than how often each of these trough types is observed. Considering the LOS00 and LOSSA lines-of-sight in Figure 4.17, we present below a comparative assessment of the expected properties of C iv00 and C ivSA troughs, relating iv these to the observational findings above (we then discuss C S0 troughs as an intermediate case). Specifically, we consider BAL-trough profile properties (e.g., depth, width, EW, velocity, and profile shape) and BAL-variability characteristics (e.g., EW and fractional EW variation strengths, depth variation profiles). Our comparative assessment points are the following: 1. The column density of outflowing gas is considerably larger along LOSSA than along LOS00, while their line-of-sight covering factors need not differ substantially. If the iv column density plays a role in setting trough depth, it is expected that C SA troughs iv will generally be deeper than C 00 troughs (observational findings 1 and 2 above). While some BAL quasars are known to have highly saturated C iv troughs with depths largely set by the line-of-sight covering factor (rather than column density; see Section 4.1), it is not clear that all C iv troughs are highly saturated. Indeed, variability studies suggest that some C iv troughs are not highly saturated (see Sec- tion 4.1). The C iv troughs with detailed previous studies showing strong saturation are C ivS0 or C ivSA troughs, while, to our knowledge, no C iv00 troughs have been demonstrated to be highly saturated. Broadly consistent with this, we note that P v absorption corresponding to C iv00 troughs is rare and weak (see Section 4.3.3). iv 2. Given the poloidal velocity field of the model, we expect C 00 troughs to have iv generally higher minimum outflow velocities than C SA troughs, as observed (see observational finding 3). This is because LOS00 intersects gas with a high outflow velocity without intersecting much gas with a low outflow velocity. LOSSA, on the other hand, primarily intersects gas with a low outflow velocity. LOSSA might also intersect gas with a high outflow velocity if such gas extends close to the accretion- disk surface (i.e., prompt acceleration) at small radii, as appears required by our 3The evidence for high global covering factors of low-ionization gas has been most notably presented for BAL quasars with detected Mg ii absorption (e.g., Boroson & Meyers 1992; Turnshek et al. 1994; Zhang et al. 2010). Although we primarily use Al iii absorption to identify lines-of- iv ii sight with low-ionization gas, 75% of the objects in our C SA sample show Mg absorption at corresponding velocities (see Section≈ 4.2). 149 !""$% !""$% Fig. 4.17: Density (top) and poloidal velocity (bottom) maps of the disk-wind model (adapted from Figure 2 of Proga et al. 2000). The SMBH is located at (0, 0). The x and y-axes show the distance from the SMBH in units of the radius at the inner edge of the disk, where r = 3 r and r is the Schwarzschild radius of a black hole. The bold ∗ × S S black arrows indicate lines-of-sight with different viewing inclinations; LOS00 and LOSSA iv iv are the two lines-of-sight along which an observer would see C 00 and C SA troughs, respectively. 150 iv results showing high vmax values for C SA troughs. These same considerations can iv iv also explain the larger velocity widths of C SA troughs compared to C 00 troughs (see observational finding 2). 3. Given that the EW of a trough is set by a combination of its depth and width, we expect from the two comparative assessment points above that C ivSA troughs will iv have larger EWs than C 00 troughs, as observed (see observational findings 1 and 2). 4. From the model we expect that Al iii BALs will be mainly formed in the region close to the disk with high density and small poloidal velocity, while C iv BALs will be formed within both high-velocity and low-velocity gas. Thus, we expect that Al iii troughs will reside within the lower velocity portions of C iv troughs, as observed (observational finding 1 and Voit et al. 1993). 5. The model shows that the optical depth is velocity-dependent, and that it is gener- ally higher for low poloidal velocities. Therefore, the low-velocity portions of C ivSA troughs that align with corresponding Al iii troughs are likely to be more saturated than the high-velocity portions, perhaps partly leading to their larger depths (ob- servational finding 1). Therefore these portions should be less variable than the high-velocity portions. This behavior is observed (observational finding 4). 6. Previous studies have presented evidence that ionization-level changes likely have a significant role in driving some BAL variability (see Section 4.1). Strongly saturated lines will be less susceptible to variability driven by ionization-level changes. Given that C ivSA troughs are likely more saturated than C iv00 troughs, they are expected to be less variable. This is observed in an absolute sense and even more strongly in a fractional sense (observational findings 4–6). 7. The model indicates that the C iv optical depth along LOSSA is substantially larger than the Al iii optical depth. Therefore, Al iii troughs are expected to be more variable than C ivSA troughs. This behavior is observed in an absolute sense and even more strongly in a fractional sense (observational finding 8). The basic expectations of the disk-wind model for the characteristics of the C iv00 and C ivSA samples show qualitative agreement with our observational findings. In this model, a line-of-sight along which an observer would see a C ivS0 trough is expected to intercept at least some less ionized gas than LOS00. Consistent with this expectation, our iv observational findings show that the C S0 sample is an intermediate case between the iv iv C 00 and C SA samples (observational findings 1–7). There are a number of ways the results above might be advanced, and here we highlight four that appear particularly promising. First, for the reasons discussed in Sec- tion 4.1, our current work has made use of the strong C iv, Si iv, and Al iii BAL transitions 151 as a basic measure of average line-of-sight ionization level. These transitions, spanning a factor of 2.5 in ionization potential, have served effectively for our work. However, this ≈ ionization-potential range could be expanded with the use of additional lower ionization (e.g., Mg ii, Fe ii, and Fe iii) and higher ionization (e.g., Ne vii, N v,O vi) transitions, thereby presumably probing wind zones even closer to and further from, respectively, the accretion disk. Second, the profiles and variability of the emission lines for large samples of BAL quasars with C iv00,C ivS0, and C ivSA troughs should be measured systematically and compared with predictions (e.g., Murray & Chiang 1997; Flohic et al. 2012). For a flattened Broad Line Region geometry, to first order one might expect the emission lines for BAL quasars with C ivSA troughs to be generally the broadest (though different emission lines, tracing different phases of the Broad Line Region, may behave differently). Third, the ongoing BOSS ancillary project and upcoming SDSS-IV Time Domain Spectroscopic Survey (TDSS)4 observations will both enlarge the sample size and improve the temporal sampling pattern for BAL quasars. This will allow variability to be used even more effec- tively as a tool for assessing correlated changes of ionization level, kinematics, and column density. Finally, while we have found generally good qualitative agreement with expecta- tions for the disk-wind model, our ability to perform quantitative comparisons has been limited by the available simulation results. Future simulations capable of predicting trough profiles and variability (e.g., Higginbottom et al. 2013; D. Proga 2013, priv. comm.) can be quantitatively tested and constrained using large-sample measurements such as those provided here. Alternatives to the disk-wind model should also be developed to the point where quantitative testing is possible. 4The current planning for SDSS-IV is briefly described at http://www.sdss3.org/future/ 152 Chapter 5 Summary In this dissertation I have studied characteristics and time variability of BAL troughs utilizing multi-epoch observations of BAL quasars from SDSS-I/II and SDSS-III BOSS. The main result of this thesis are following: 1. In Chapter 2, I present 21 examples of C iv BAL trough disappearance in 19 quasars selected from systematic multi-epoch observations of 582 bright BAL quasars (1.9 < z < 4.5) by the SDSS-I/II and SDSS-III. The observations span 1.1–3.9 yr rest-frame timescales, longer than have been sampled in many previous BAL variability studies. On these timescales, 2.3% of C iv BAL troughs disappear and 3.3% of BAL ≈ ≈ quasars show a disappearing trough. These observed frequencies suggest that many C iv BAL absorbers spend on average at most a century along our line of sight to their quasar. Ten of the 19 BAL quasars showing C iv BAL disappearance have apparently transformed from BAL to non-BAL quasars; these are the first reported examples of such transformations. The BAL troughs that disappear tend to be those with small- to-moderate equivalent widths, relatively shallow depths, and high outflow velocities. Other non-disappearing C iv BALs in those nine objects having multiple troughs tend to weaken when one of them disappears, indicating a connection between the disappearing and non-disappearing troughs, even for velocity separations as large as 10000–15000 km s−1. I discuss possible origins of this connection including disk-wind rotation and changes in shielding gas. 2. In Chapter 3, I present a review study on variation characteristics for C iv and Si iv broad absorption line (BAL) troughs utilizing 699 high-quality spectra of 291 z 2 ≥ quasars observed on rest-frame timescales between 6 hours to 3.7 yr. We identify 428 distinct C iv BAL troughs and 291 distinct Si iv BAL troughs in our sample where about 50–60% of them detectably vary on rest-frame timescales of 1–3.7 yr. Varia- tions usually occur in relatively narrow portions of BAL troughs and these portions typically span < 30% of the trough velocity width. The EW and fractional EW variations of C iv and Si iv BAL troughs increase with sampled rest-frame timescale, and the measurements for fractional EW variations suggest an average lifetime for a BAL trough along our line-of-sight of a few thousand years. The distribution of EW variations are symmetric and non-Gaussian both for C iv and Si iv. The distribu- tions of EW and fractional EW variations indicate a connection to disappearance and emergence which are extreme events of the continuum of BAL variability. A simple 153 one-dimensional random-walk model can explain the evolution of BAL-trough EWs. There are strong correlations between the EW and fractional EW variations of C iv and Si iv BAL troughs. The multiple BAL troughs of each transition show a strong correlated variation. The multiple troughs of the same ion usually strengthen or weaken together and their EW variations are strongly correlated for velocity offsets as large as 15000–20000 km s−1. I also characterize the EW variations as a function of quasar properties such as bolometric luminosity, Eddington luminosity ratio, black hole mass, redshift, and the radio loudness parameter, R, however find no significant correlation between EW variations and these quasar properties. 3. In Chapter 4, I investigate how the profile and multi-year variability properties of a large sample of C iv BAL troughs change when BALs from Si iv and/or Al iii are present at corresponding velocities, indicating that the line-of-sight intercepts at least some lower ionization gas. We derive a number of observational results for C iv BALs separated according to the presence or absence of accompanying lower ionization transitions, including measurements of composite profile shapes, EW, characteristic velocities, composite variation profiles, and EW variability. We also measure the cor- relations between EW and fractional-EW variability for C iv, Si iv, and Al iii. Our measurements reveal the basic correlated changes between ionization level, kinemat- ics, and column density expected in accretion-disk wind models as the line-of-sight is varied; e.g., lines-of-sight including lower ionization material generally show deeper and broader C iv troughs that have smaller minimum velocities and that are less variable. Many C iv BALs with no accompanying Si iv or Al iii BALs may have only mild or no saturation. 154 Bibliography Abazajian, K. N., Adelman-McCarthy, J. K., Ag¨ueros, M. A., et al. 2009, ApJS, 182, 543 Aihara, H., Allende Prieto, C., An, D., et al. 2011, ApJS, 193, 29 Allen, J. T., Hewett, P. C., Maddox, N., Richards, G. T., & Belokurov, V. 2011, MNRAS, 410, 860 Anderson, L., Aubourg, E., Bailey, S., et al. 2012, MNRAS, 427, 3435 Arav, N. 1997, in Astronomical Society of the Pacific Conference Series, Vol. 128, Mass Ejection from Active Galactic Nuclei, ed. N. Arav, I. Shlosman, & R. J. Weymann, 208 Arav, N., Becker, R. H., Laurent-Muehleisen, S. A., et al. 1999b, ApJ, 524, 566 Arav, N., Borguet, B., Chamberlain, C., Edmonds, D., & Danforth, C. 2013, ArXiv e-prints Arav, N., Korista, K. T., de Kool, M., Junkkarinen, V. T., & Begelman, M. C. 1999a, ApJ, 516, 27 Arav, N., de Kool, M., Korista, K. T., et al. 2001, ApJ, 561, 118 Barlow, T. A. 1993, PhD thesis, UCSD Barlow, T. A., Junkkarinen, V. T., Burbidge, E. M., et al. 1992, ApJ, 397, 81 Baskin, A., Laor, A., & Hamann, F. 2013, MNRAS, 432, 1525 Becker, R. H., White, R. L., Gregg, M. D., et al. 2000, ApJ, 538, 72 Becker, R. H., White, R. L., & Helfand, D. J. 1995, ApJ, 450, 559 Bentz, M. C., Peterson, B. M., Netzer, H., Pogge, R. W., & Vestergaard, M. 2009, ApJ, 697, 160 Bevington, P. R., & Robinson, D. K. 2003, Data reduction and error analysis for the physical sciences (McGraw-Hill, Inc.) Blandford, R. D., & Payne, D. G. 1982, MNRAS, 199, 883 Bolton, A. S., Burles, S., Schlegel, D. J., Eisenstein, D. J., & Brinkmann, J. 2004, AJ, 127, 1860 Bolton, A. S., Schlegel, D. J., Aubourg, E.,´ et al. 2012, AJ, 144, 144 155 Borguet, B. C. J., Arav, N., Edmonds, D., Chamberlain, C., & Benn, C. 2013, ApJ, 762, 49 Borguet, B. C. J., Edmonds, D., Arav, N., Benn, C., & Chamberlain, C. 2012, ApJ, 758, 69 Boroson, T. A., & Meyers, K. A. 1992, ApJ, 397, 442 Brotherton, M. S., van Breugel, W., Smith, R. J., et al. 1998, ApJ, 505, L7 Busca, N. G., Delubac, T., Rich, J., et al. 2013, A&A, 552, A96 Capellupo, D. M., Hamann, F., Shields, J. C., Halpern, J. P., & Barlow, T. A. 2013, MNRAS, 429, 1872 Capellupo, D. M., Hamann, F., Shields, J. C., Rodr´ıguez Hidalgo, P., & Barlow, T. A. 2011, MNRAS, 413, 908 —. 2012, MNRAS, 422, 3249 Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245 Chartas, G., Saez, C., Brandt, W. N., Giustini, M., & Garmire, G. P. 2009, ApJ, 706, 644 Collin, S., Kawaguchi, T., Peterson, B. M., & Vestergaard, M. 2006, A&A, 456, 75 Condon, J. J., Cotton, W. D., Greisen, E. W., et al. 1998, AJ, 115, 1693 Crenshaw, D. M., Kraemer, S. B., & George, I. M. 2003, ARA&A, 41, 117 Dawson, K. S., Schlegel, D. J., Ahn, C. P., et al. 2013, AJ, 145, 10 Di Matteo, T., Springel, V., & Hernquist, L. 2005, Nature, 433, 604 DiPompeo, M. A., Brotherton, M. S., & De Breuck, C. 2012, ApJ, 752, 6 Eisenstein, D. J., Weinberg, D. H., Agol, E., et al. 2011, AJ, 142, 72 Emmering, R. T., Blandford, R. D., & Shlosman, I. 1992, ApJ, 385, 460 Faucher-Gigu`ere, C.-A., Quataert, E., & Murray, N. 2012, MNRAS, 420, 1347 Filiz Ak, N., Brandt, W. N., Hall, P. B., et al. 2012, ApJ, 757, 114 —. 2013, ArXiv e-prints Gallagher, S. C., Brandt, W. N., Chartas, G., & Garmire, G. P. 2002, ApJ, 567, 37 Gallagher, S. C., Brandt, W. N., Chartas, G., et al. 2006, ApJ, 644, 709 156 Gallagher, S. C., Brandt, W. N., Wills, B. J., et al. 2004, ApJ, 603, 425 Ganguly, R., & Brotherton, M. S. 2008, ApJ, 672, 102 Ganguly, R., Brotherton, M. S., Cales, S., et al. 2007, ApJ, 665, 990 Gehrels, N. 1986, ApJ, 303, 336 Gibson, R. R., Brandt, W. N., Gallagher, S. C., Hewett, P. C., & Schneider, D. P. 2010, ApJ, 713, 220 Gibson, R. R., Brandt, W. N., Schneider, D. P., & Gallagher, S. C. 2008, ApJ, 675, 985 Gibson, R. R., Jiang, L., Brandt, W. N., et al. 2009, ApJ, 692, 758 Gunn, J. E., Carr, M., Rockosi, C., et al. 1998, AJ, 116, 3040 Gunn, J. E., Siegmund, W. A., Mannery, E. J., et al. 2006, AJ, 131, 2332 Haggard, D., Arraki, K. S., Green, P. J., Aldcroft, T., & Anderson, S. F. 2012, in Astro- nomical Society of the Pacific Conference Series, Vol. 460, AGN Winds in Charleston, ed. G. Chartas, F. Hamann, & K. M. Leighly, 98 Hall, P. B., Anosov, K., White, R. L., et al. 2011, MNRAS, 411, 2653 Hall, P. B., Anderson, S. F., Strauss, M. A., et al. 2002, ApJS, 141, 267 Hamann, F. 1998, ApJ, 500, 798 Hamann, F., Kaplan, K. F., Rodr´ıguez Hidalgo, P., Prochaska, J. X., & Herbert-Fort, S. 2008, MNRAS, 391, L39 Hamann, F., & Sabra, B. 2004, in Astronomical Society of the Pacific Conference Series, Vol. 311, AGN Physics with the Sloan Digital Sky Survey, ed. G. T. Richards & P. B. Hall, 203 Hamann, F. W., Netzer, H., & Shields, J. C. 2000, ApJ, 536, 101 Hewett, P. C., Foltz, C. B., & Chaffee, F. H. 1995, AJ, 109, 1498 Hewett, P. C., & Wild, V. 2010, MNRAS, 405, 2302 Higginbottom, N., Knigge, C., Long, K. S., Sim, S. A., & Matthews, J. H. 2013, MNRAS, 436, 1390 Hopkins, P. F., Strauss, M. A., Hall, P. B., et al. 2004, AJ, 128, 1112 Junkkarinen, V., Shields, G. A., Beaver, E. A., et al. 2001, ApJ, 549, L155 157 Kaspi, S., Brandt, W. N., Maoz, D., et al. 2007, ApJ, 659, 997 Kaspi, S., Maoz, D., Netzer, H., et al. 2005, ApJ, 629, 61 Kaspi, S., Brandt, W. N., George, I. M., et al. 2002, ApJ, 574, 643 Kelly, B. C. 2007, ApJ, 665, 1489 King, A. R. 2010, MNRAS, 402, 1516 Konigl, A., & Kartje, J. F. 1994, ApJ, 434, 446 Krongold, Y., Binette, L., & Hern´andez-Ibarra, F. 2010, ApJ, 724, L203 Laor, A., & Brandt, W. N. 2002, ApJ, 569, 641 Leighly, K. M. 2004, ApJ, 611, 125 Leighly, K. M., Hamann, F., Casebeer, D. A., & Grupe, D. 2009, ApJ, 701, 176 Lilliefors, H. 1967, Journal of the American Statistical Association, 62, 399 Lucy, A. B., Leighly, K. M., Terndrup, D. M., Dietrich, M., & Gallagher, S. C. 2014, ArXiv e-prints Lundgren, B. F., Wilhite, B. C., Brunner, R. J., et al. 2007, ApJ, 656, 73 Lundgren, B. F., & York, D. G. e. a. 2014 in preparation Ma, F. 2002, MNRAS, 335, L99 Margala, D., & Kirkby, D. 2011, in SDSS BOSS QSO Workshop, Princeton, NJ, USA Merloni, A., Predehl, P., Becker, W., et al. 2012, ArXiv e-prints Miller, B. P., Brandt, W. N., Gibson, R. R., Garmire, G. P., & Shemmer, O. 2009, ApJ, 702, 911 Miller, B. P., Welling, C. A., Brandt, W. N., & Gibson, R. R. 2012, in Astronomical Society of the Pacific Conference Series, Vol. 460, AGN Winds in Charleston, ed. G. Chartas, F. Hamann, & K. M. Leighly, 118 Misawa, T., Eracleous, M., Charlton, J. C., & Kashikawa, N. 2007, ApJ, 660, 152 Murray, N., Chiang, J., Grossman, S. A., & Voit, G. M. 1995, ApJ, 451, 498 O’Brien, P. T., Reeves, J. N., Simpson, C., & Ward, M. J. 2005, MNRAS, 360, L25 Ogle, P. M., Cohen, M. H., Miller, J. S., et al. 1999, ApJS, 125, 1 158 Pˆaris, I., Petitjean, P., Aubourg, E.,´ et al. 2012, A&A, 548, A66 —. 2013, ArXiv e-prints Pei, Y. C. 1992, ApJ, 395, 130 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) Proga, D., & Kallman, T. R. 2004, ApJ, 616, 688 Proga, D., Kallman, T. R., Drew, J. E., & Hartley, L. E. 2002, ApJ, 572, 382 Proga, D., Stone, J. M., & Kallman, T. R. 2000, ApJ, 543, 686 Randles, R. H., Fligner, M. A., Policello, G. E., & Wolfe, D. A. 1980, Journal of American Statistic Association, 75, 168 Reichard, T. A., Richards, G. T., Hall, P. B., et al. 2003, AJ, 126, 2594 Richards, G. T., Fan, X., Newberg, H. J., et al. 2002, AJ, 123, 2945 Richards, G. T., Kruczek, N. E., Gallagher, S. C., et al. 2011, AJ, 141, 167 Ross, N. P., Myers, A. D., Sheldon, E. S., et al. 2012, ApJS, 199, 3 Rupke, D. S. N., & Veilleux, S. 2011, ApJ, 729, L27 Saez, C., Brandt, W. N., Gallagher, S. C., Bauer, F. E., & Garmire, G. P. 2012, ApJ, 759, 42 Savitzky, A., & Golay, M. J. E. 1964, Analytical Chemistry, 36, 1627 Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525 Schneider, D. P., Hall, P. B., Richards, G. T., et al. 2007, AJ, 134, 102 Schneider, D. P., Richards, G. T., Hall, P. B., et al. 2010, AJ, 139, 2360 Shankar, F., Dai, X., & Sivakoff, G. R. 2008, ApJ, 687, 859 Shen, Y. 2013, Bulletin of the Astronomical Society of India, 41, 61 Shen, Y., Greene, J. E., Strauss, M. A., Richards, G. T., & Schneider, D. P. 2008, ApJ, 680, 169 Shen, Y., Richards, G. T., Strauss, M. A., et al. 2011, ApJS, 194, 45 Sim, S. A., Proga, D., Kurosawa, R., et al. 2012, MNRAS, 426, 2859 159 Sim, S. A., Proga, D., Miller, L., Long, K. S., & Turner, T. J. 2010, MNRAS, 408, 1396 Slosar, A., Irˇsiˇc, V., Kirkby, D., et al. 2013, Journal of Cosmology and Astroparticle Physics, 4, 26 Smee, S. A., Gunn, J. E., Uomoto, A., et al. 2013, AJ, 146, 32 Smith, L. J., & Penston, M. V. 1988, MNRAS, 235, 551 Spergel, D. N., Verde, L., Peiris, H. V., et al. 2003, ApJS, 148, 175 Springel, V., Di Matteo, T., & Hernquist, L. 2005, MNRAS, 361, 776 Sturm, E., Gonz´alez-Alfonso, E., Veilleux, S., et al. 2011, ApJ, 733, L16 Suzuki, N. 2013, in SDSS-III Collaboration Meeting, Baltimore, MD, USA Telfer, R. C., Kriss, G. A., Zheng, W., Davidsen, A. F., & Green, R. F. 1998, ApJ, 509, 132 Trump, J. R., Hall, P. B., Reichard, T. A., et al. 2006, ApJS, 165, 1 Turnshek, D. A. 1988, in QSO Absorption Lines: Probing the Universe, ed. J. C. Blades, D. A. Turnshek, & C. A. Norman, 17 Turnshek, D. A., Grillmair, C. J., Foltz, C. B., & Weymann, R. J. 1988, ApJ, 325, 651 Turnshek, D. A., Espey, B. R., Kopko, Jr., M., et al. 1994, ApJ, 428, 93 Vanden Berk, D. E., Wilhite, B. C., Kron, R. G., et al. 2004, ApJ, 601, 692 Vilkoviskij, E. Y., & Irwin, M. J. 2001, MNRAS, 321, 4 Vivek, M., Srianand, R., Mahabal, A., & Kuriakose, V. C. 2012, MNRAS, 421, L107 Voit, G. M., Weymann, R. J., & Korista, K. T. 1993, ApJ, 413, 95 Welling, C. A., Miller, B. P., Brandt, W. N., Capellupo, D. M., & Gibson, R. R. 2013, MNRAS, submitted Weymann, R. J., Carswell, R. F., & Smith, M. G. 1981, ARA&A, 19, 41 Weymann, R. J., Morris, S. L., Foltz, C. B., & Hewett, P. C. 1991, ApJ, 373, 23 White, R. L., Becker, R. H., Gregg, M. D., et al. 2000, ApJS, 126, 133 Wildy, C., Goad, M. R., & Allen, J. T. 2013, MNRAS York, D. G., Adelman, J., Anderson, Jr., J. E., et al. 2000, AJ, 120, 1579 Zhang, S., Wang, T.-G., Wang, H., et al. 2010, ApJ, 714, 367 Vita Nurten Filiz Ak Education The Pennsylvania State University State College, Pennsylvania 2010–Present Ph.D. in Astronomy & Astrophysics, expected in May 2014 Area of Specialization: Broad Absorption Line Quasars Canakkale 18 Mart University Canakkale, Turkey 2003–2006 M.S. in Physics Ankara University Ankara, Turkey 1999–2003 B.S. in Astronomy and Space Science Awards and Honors Young SDSS Astronomer Travel Assistance Award 2013 Zaccheus Daniel Foundation for Astronomical Science Grant 2012, 2013 Turkey Higher Education Council Research Abroad Award 2009–2010 Erciyes University Research Fellowship for Science Students 2007–2010 TUBITAK Graduate Student Research Program Fellowship 2005–2006 Research Experience Doctoral Research The Pennsylvania State University 2009–Present Thesis Advisor: Prof. William N. Brandt I studied BAL variability in quasar UV spectra obtained by SDSS-I/II and III. Graduate Research Canakkale 18 Mart University 2005–2009 Research Advisor: Prof. Zeki Eker I studied spectroscopic and photometric analysis of chromospherically active binary star systems. Graduate Research Canakkale 18 Mart University 2003–2005 Research Advisor: Prof. Osman Demircan I studied stellar rotation and its effect on chromospherical activity. Teaching Experience Teaching Assistant The Pennsylvania State University 2010–2011 I taught telescope labs and assisted lectures on several introductory astronomy classes. Guest Lecturer Erciyes University 2007–2009 I prepared and taught lectures on Introduction to Spectroscopy class for four semesters. Teaching Assistant Canakkale 18 Mart University, Canakkale, Turkey 2003–2006 I assisted lectures on several introductory astronomy and physics classes.