A SEARCH FOR SUBSTELLAR COMPANIONS AROUND PRE-MAIN SEQUENCE

STARS USING INFRARED SPECTROSCOPY

A DISSERTATION SUBMITTED TO THE GRADUATE DIVISION OF THE UNIVERSITY OF HAWAI‘I AT MANOA¯ IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

IN

ASTRONOMY

September 2020

By Larissa A. Nofi

Dissertation Committee:

Daniel Huber, Chairperson Klaus Hodapp Andrew Howard Paul Lucey Karen Meech John Rayner c Copyright 2020 by Larissa A. Nofi All Rights Reserved

ii This dissertation is dedicated to all those who had confidence in my abilities. Thank you for the support.

iii Acknowledgements

First and foremost, I would like to acknowledge the mentors and colleagues who supported this work. I am grateful to my advisor, Dan Huber, and my committee for valuable insights and guidance. I wish to express my thanks to the IGRINS team and other collaborators from multiple institutions, all of whom contributed meaningfully to this dissertation. Thank you to my fellow graduate students for your friendship and support, and particularly to Kelly

Blumenthal, for being a constant presence and source of motivation. I am so grateful we were able to share every step of this process together.

I would also like to acknowledge the efforts of the Lowell Discovery Telescope staff and the use of the Immersion Grating Infrared Spectrometer (IGRINS), developed by a collaboration between University of Texas at Austin and the Korea Astronomy and Space

Science Institute (KASI). I acknowledge the gracious support of the Lowell Pre-doctoral

Fellowship by the BF Foundation, funding provided by the Visiting Astronomer at the

Infrared Telescope Facility program, which is operated by the University of Hawaii under contract 80HGTR19D0030 with the National Aeronautics and Space Administration, and the Institute for Astronomy and University of Hawaii for supporting a graduate scholarship.

I am grateful to my long-time mentor, Sloane Wiktorowicz, for consistently inspiring and challenging me, and for being a model of a scientist I wished to become. Thank you to colleagues at Lick Observatory and UC Santa Cruz who encouraged me in my education prior to graduate school, and to the professors at City College of San Francisco who saw potential in me before I was able to see it in myself. This accomplishment belongs to all of you.

iv I would particularly like to acknowledge and thank my husband, who never wavered in his support, and who has now moved across an ocean three times with me so that I could make this ambition a reality.

v Abstract

Observing and characterizing newly-formed planets around young stars is important for developing planet formation and evolution theory. However, given the challenges involved in detecting young planetary systems, current models are primarily based on observed systems that are billions of years old. In this dissertation, I present a survey to characterize pre-main sequence stars, and detect or confirm young substellar companions using the radial velocity

(RV) method. This survey used the Immersion Grating Infrared Spectrograph (IGRINS), which was deployed as a visiting instrument at the 4.3-m Lowell Discovery Telescope (LDT).

Infrared spectroscopy is preferable to optical because it is less affected by RV variability triggered by stellar activity effects, such as starspots. The initial survey consisted of 70 pre-main sequence stars with ages <5 Myr in the relatively nearby Taurus star-forming region. I measured the projected rotational velocity (v sin i) of these pre-main sequence stars, and related the rotation (and other properties) to young star evolution. Additionally,

I investigated a subset of 10 targets to look for substellar companions. The RV survey is generally sensitive to hot Jupiters with mass >6 MJ and brown dwarfs within ∼1 AU. No companions were detected in this sample, however, upper limits on occurrence rates

were estimated based on injection/recovery simulations, and are consistent with multiple

theories of formation and evolution. Correlations between stellar properties and RV scatter

indicate that rotation, stellar activity, and disk accretion are possible contributing factors

to RV variability. The average infrared RV scatter is roughly half the typical optical RV

scatter for very young stars, confirming the potential of infrared RV surveys to detect young

substellar companions.

vi Table of Contents

Acknowledgements ...... iv

Abstract ...... vi

List of Tables ...... ix

List of Figures ...... x

Chapter 1: Introduction ...... 1

1.1 Motivation ...... 1

1.2 Exoplanet Detection Techniques ...... 5

1.2.1 Direct Imaging Method ...... 5

1.2.2 Transit Method ...... 6

1.2.3 Radial Velocity Method ...... 8

1.3 Substellar Companions Around Young Stars ...... 12

1.3.1 Formation and Evolution Theory ...... 12

1.3.2 Occurrence Rates of Substellar Companions ...... 24

1.3.3 Young Substellar Candidate Discoveries ...... 29

1.4 Science Objectives ...... 33

Chapter 2: Survey Overview ...... 49

2.1 Sample Selection ...... 49

2.2 Infrared Spectroscopy Survey ...... 56

2.2.1 Finding Young Substellar Companions: Challenges and Solutions . . 56

2.2.2 Observations and Data Reduction ...... 62

vii 2.3 Prior Optical Spectroscopy Survey ...... 67

2.4 Optical Photometric Survey ...... 73

Chapter 3: Stellar Properties of Pre-Main Sequence Stars ...... 84

3.1 Introduction ...... 85

3.2 Analysis ...... 87

3.2.1 Stellar Effective Temperatures ...... 87

3.2.2 Infrared v sin i ...... 88

3.2.3 Optical v sin i ...... 95

3.2.4 Stellar Rotation Periods ...... 97

3.2.5 Stellar Radius Limits ...... 98

3.3 Discussion ...... 98

3.3.1 Distributions of Stellar Parameters ...... 98

3.3.2 Effects of Multiplicity on v sin i ...... 100

3.3.3 Comparison of Optical and Infrared v sin i ...... 103

3.3.4 Correlation of v sin i and Evolutionary States ...... 109

3.4 Conclusions ...... 115

Chapter 4: A Search for Substellar Companions around Pre-Main Sequence Stars . . 128

4.1 Methodology ...... 129

4.1.1 Forward Modeling Technique ...... 129

4.1.2 RV Measurements ...... 131

4.1.3 RV Precision ...... 142

4.1.4 Periodogram Analysis ...... 148

4.1.5 Sources of Spurious RV Signals ...... 153

4.1.6 Detection Limits from Injection and Recovery Tests ...... 158

4.2 Results and Discussion ...... 160

4.2.1 Individual System Results ...... 160

4.2.2 Discussion ...... 196

Chapter 5: Conclusions and Summary ...... 228

viii Appendix: Radial Velocities of Pre-Main Sequence Stars and the RV Standard Star 236

ix List of Tables

2.1 Sample of 70 Pre-Main Sequence Stars ...... 53

2.2 IGRINS Specifications ...... 63

2.3 Observations Summary ...... 65

3.1 K-S and A-D Test Results ...... 100

3.2 Stellar Properties of Pre-Main Sequence Stars ...... 117

4.1 RV Target Properties and Observations ...... 163

4.2 Completeness of Individual RV Target Datasets ...... 204

4.3 Infrared RV Scatter ...... 212

A.1 GJ 281 RVs ...... 236

A.2 CI Tau RVs ...... 239

A.3 V830 Tau RVs ...... 242

A.4 DK Tau RVs ...... 246

A.5 V1075 Tau RVs ...... 249

A.6 AA Tau RVs ...... 253

A.7 DM Tau RVs ...... 255

A.8 GI Tau RVs ...... 256

A.9 GM Tau RVs ...... 257

A.10 IQ Tau RVs ...... 258

A.11 LkCa 15 RVs ...... 260

x List of Figures

1.1 The Brown Dwarf Desert ...... 4

1.2 Example of Direct Imaging Detection ...... 6

1.3 Example Transit Detection ...... 7

1.4 Example RV Detection ...... 9

1.5 Comparison of Orbital Eccentricity for Hot and Warm Jupiters ...... 19

1.6 Formation Scenarios for Hot and Warm Jupiters ...... 21

1.7 Occurrence Rates of Hot and Warm Jupiters ...... 28

2.1 K band Magnitudes of the 70 Pre-main Sequence Stars in the Sample . . . 51

2.2 Color-Magnitude Diagram of Sample ...... 57

2.3 Hertzsprung-Russell Diagram of Sample ...... 58

2.4 Diagram Illustrating How Starspots Cause RV Variability ...... 59

2.5 RVs Measured in Optical and Infrared Wavelengths ...... 61

2.6 IGRINS Echellogram ...... 64

2.7 IGRINS Echellograms Before and After Wavelength Calibration and

Distortion Correction...... 66

2.8 IGRINS Example Spectra ...... 68

2.9 Example of Line Bisector Analysis ...... 70

2.10 Detection of a Young Hot Jupiter Candidate, CI Tau b ...... 72

2.11 Example Lightcurve ...... 75

3.1 Example Spectra with Various v sin i Estimates ...... 89

xi 3.2 Relations Between the FWHM of the CCF and v sin i ...... 91

3.3 Comparison of v sin i Measurements Using Two Different Methods . . . . . 94

3.4 Comparison of Stellar Properties ...... 99

3.5 The v sin i Distribution of Single and Multiple System Pre-Main Sequence

Stars ...... 102

3.6 Comparison of Optical and Infrared v sin i Measurements ...... 105

3.7 Comparison of Infrared and Literature v sin i Measurements ...... 108

3.8 Comparison of Infrared and Literature v sin i Measurements Based on Stellar

Classification ...... 109

3.9 The v sin i Distributions for CTTSs and WTTSs ...... 111

3.10 H-R Diagram and v sin i Correlation ...... 114

4.1 Forward Modeling Technique ...... 130

4.2 Comparison of the CSHELL Program and the Independent RV Program . . 137

4.3 The Modified Spectral Region Used in the Analysis ...... 140

4.4 The Improved Continuum Fit ...... 141

4.5 RVs of the Standard Star, GJ 281 ...... 143

4.6 Estimated Detection Limits for Various Precisions ...... 147

4.7 Example of a Lomb-Scargle Periodogram ...... 150

4.8 Example RV Phase Curve ...... 152

4.9 Example of a Spectral Window Function ...... 154

4.10 Technique to Test for Aliased Signals ...... 156

4.11 Example of a Gaussian Process Fit ...... 158

4.12 Example of a Completeness Contour Plot ...... 161

4.13 The CI Tau Lightcurve from K2 ...... 166

4.14 CI Tau RVs vs. JD ...... 168

4.15 CI Tau Periodogram Analysis ...... 170

4.16 CI Tau Completeness Contour Plot ...... 171

xii 4.17 V830 Tau RV Analysis ...... 174

4.18 The V830 Tau Lightcurve from K2 ...... 175

4.19 V830 Tau Completeness Contour Plot ...... 177

4.20 The DK Tau Lightcurve from K2 ...... 178

4.21 DK Tau RV Analysis ...... 180

4.22 DK Tau Doublet ...... 182

4.23 DK Tau Completeness Contour ...... 183

4.24 The V1075 Tau Lightcurve from K2 ...... 184

4.25 V1075 Tau RV Analysis ...... 185

4.26 V1075 Tau Completeness Contour Plot ...... 186

4.27 The GM Tau Lightcurve from K2 ...... 188

4.28 The IQ Tau Lightcurve from K2 ...... 188

4.29 The LkCa 15 Lightcurve from K2 ...... 188

4.30 AA Tau RV Analysis ...... 190

4.31 DM Tau RV Analysis ...... 191

4.32 GI Tau RV Analysis ...... 192

4.33 GM Tau RV Analysis ...... 193

4.34 IQ Tau RV Analysis ...... 194

4.35 LkCa 15 RV Analysis ...... 195

4.36 AA Tau Completeness Contour Plot ...... 197

4.37 DM Tau Completeness Contour Plot ...... 198

4.38 GI Tau Completeness Contour Plot ...... 199

4.39 GM Tau Completeness Contour Plot ...... 200

4.40 IQ Tau Completeness Contour Plot ...... 201

4.41 LkCa 15 Completeness Contour Plot ...... 202

4.42 GM Tau and IQ Tau Completeness Contour Plots with Varied Parameters . 203

4.43 Upper Limits on Occurrence Rates for Young Substellar Companions . . . . 207

4.44 Histogram of RV Scatter ...... 211

xiii 4.45 IQ Tau Photometry ...... 214

4.46 H-R Diagram and RV Scatter Correlation ...... 215

xiv Chapter 1

Introduction

1.1 Motivation

The discovery of exoplanets over the last ∼25 years has dramatically changed our perspective on the variety of planetary systems which exist in the universe. Diverse populations, such as gas giant planets in short-period orbits, systems of tightly-packed planets, and relatively common super-Earths have challenged formation and evolution theory.

One of the most intriguing classes of planets found to date are hot Jupiters. Hot Jupiters are giant planets with mass >0.3 Jupiter mass (MJ) that orbit within 0.1 au of their host star, with an orbital period <10 days. The name comes from the high equilibrium temperature of these planets, resulting from their close proximity to their host star, which can reach values close to 1500 K at a separation of 0.05 au from a solar-type star (Udry &

Santos 2007). Hot Jupiters were among the first exoplanets discovered (Mayor & Queloz

1995), and are easy to detect with common detection methods because of their high mass and close orbits.

Warm Jupiters are similar to hot Jupiters but are found at wider separations from the host star. They are gas giant planets orbiting in the region from 0.1-1 au with an orbital period between 10 and 100 days. While warm Jupiters have wider separations from the host star, their high masses also make them relatively easy to detect using common techniques,

1 in comparison to low-mass terrestrial planets or gas and ice giants in larger orbits similar to that of Jupiter.

Hot and warm Jupiter discoveries are important because they offer a view of planet formation and evolution unseen in our own solar system. They are thought to have a critical impact on the early architecture of systems and can provide insights into the inner system where terrestrial planets form. Long thought to be destructive to terrestrial planets

(Gonzalez et al. 2001; Lineweaver 2001), close-in gas giants may actually have very little influence on these systems (Raymond et al. 2005; Agnew et al. 2017; Georgakarakos et al.

2018), and may even draw water into the inner region to create water-rich terrestrial planets within the habitable zone (Raymond et al. 2006; Fogg & Nelson 2007; Darriba et al. 2017;

Raymond & Izidoro 2017). Furthermore, hot and warm Jupiters are of interest as potential hosts to exomoons. It is unknown if exomoons can form or survive in the conditions related to short-period gas giant planet formation and evolution (Barnes & O’Brien 2002; Gong et al. 2013; Heller & Pudritz 2015; Alvarado-Montes et al. 2017), and exomoons orbiting gas giants in the inner system could potentially reside within the habitable zone (Heller et al. 2014).

The inner system is also a significant region to study brown dwarfs. While the boundary between brown dwarfs and high-mass planets or low-mass stars is not completely distinct, a common definition for a brown dwarf is a substellar object with mass between ∼13-80 MJ (Burrows et al. 2001). This encompasses the range of objects that are massive enough to sustain deuterium fusion (differentiating them from planets) but not massive enough to fuse hydrogen in their cores (differentiating them from stars). There is an overlap of the brown dwarf and massive planet regimes in the interval ∼10–20 MJ, making it difficult to accurately classify objects in this mass range. It is thought that brown dwarfs form much like stars by fragmentation of a protostellar cloud, while planets are likely built up from protostellar disk material (Adams et al. 1989; Bate et al. 2003; Schlaufman 2018). Additional information on the formation and evolution of these objects is therefore necessary to differentiate between them (Udry & Santos 2007).

2 As surveys discovered more close-in planetary and stellar companions, it became apparent that there was a deficit of brown dwarf companions in short-period orbits around solar-mass stars. This “brown dwarf desert” is generally thought to exist within a semi- major axis .3 au, or equivalently an orbital period .5 yr (Marcy & Butler 2000; Armitage & Bonnell 2002; Grether & Lineweaver 2006). Figure 1.1 illustrates the brown dwarf desert in companion mass and orbital period (or semi-major axis). This figure clearly shows a deficit of brown dwarf companions at P<5 yr. Ma & Ge (2014) argue that the largest

deficit of objects in the brown dwarf desert are those with masses between 35–55 MJ and periods shorter than 100 days.

Several studies have demonstrated a high frequency of field brown dwarfs (Kirkpatrick et al. 1999, 2000), and brown dwarfs in wide binaries (Gizis et al. 2001), young clusters

(Mart´ınet al. 2000), and star-forming regions (B´ejaret al. 2001). Despite this apparent brown dwarf abundance, a difference of at least 1.5 orders of magnitude was measured between the number of brown dwarfs found in stellar clusters and those found in close orbits around solar-type stars (Grether & Lineweaver 2006). The brown dwarf desert is not completely dry, however; some surveys have detected short-period brown dwarfs (Marcy et al. 2001; Carmichael et al. 2019), so theory must explain a substantial, rather than complete, depletion of this population, assuming that these objects form to begin with.

The study of young stars (<5 Myr old) is important for understanding hot/warm Jupiter and brown dwarf formation. An observational study of the inner star system can provide insights into the origin of the brown dwarf desert (Huerta et al. 2008; Mahmud et al. 2011), as well as the origin of hot and warm Jupiters. It is not known if the paucity of brown dwarfs in this region is the result of formation processes or evolution. It is also not known exactly how hot Jupiters form and evolve. A survey searching for young substellar objects can significantly contribute to these questions by providing a sample of objects to compare to formation and evolution models.

This dissertation describes an observational survey of pre-main sequence stars to detect and characterize young substellar companions, such as the hot/warm Jupiters and brown

3 Figure 1.1 The brown dwarf desert in mass and period from Grether & Lineweaver (2006). Detections of companions to Sun-like stars for stellar hosts within 25 pc are indicated by large symbols, while those for host stars between 25 and 50 pc are indicated by small symbols. The stellar (open circles), brown dwarf (gray circles) and planetary (filled circles) companions are separated by dashed lines at the hydrogen (80 MJ) and deuterium (13 MJ) fusion onset masses. This figure clearly shows the brown dwarf desert for the P<5 yr companions. Detection limits are also indicated and shaded in gray, where the label “being detected” indicates regions where not all companions will be found given the survey sensitivity. The thick solid rectangle encompasses the sample that is considered less biased in relation to the remaining regions.

4 dwarfs. The remainder of this chapter will introduce common detection techniques used to identify young substellar companions, with a discussion of the challenges and strengths of each (Section 1.2), provide additional motivation and background for studying young substellar companions (Section 1.3) to 1) inform the theory of hot/warm Jupiter formation and evolution and the origin of the brown dwarf desert (Section 1.3.1) and 2) to inform planet population statistics (Section 1.3.2), and present recent young exoplanet detections and how they have contributed to formation theory (Section 1.3.3). Section 1.4 outlines the science objectives of this survey. Chapter 2 will give an overview of the survey design, including a discussion of the infrared and optical spectroscopy, and photometry used in the analysis; Chapter 3 will present characterization of pre-main sequence stars (projected rotational velocities, effective temperatures, rotation periods, and limits on radii); Chapter 4 will discuss the RV measurements and survey precision, as well as the RV analysis, results, and discussion of sensitivity, upper limits on occurrence rates, and infrared RV scatter;

Chapter 5 will present the conclusions.

1.2 Exoplanet Detection Techniques

1.2.1 Direct Imaging Method

Young substellar companions are commonly discovered using the direct imaging technique.

Because substellar objects are brightest immediately after formation, young planets are more readily detected using this technique than their mature counterparts. This method directly measures the intrinsic luminosity of substellar companions at infrared wavelengths, where the star-planet brightness contrast is lower. Direct imaging is technically challenging: the angular separation between a very faint planetary object and a much brighter host star is small. Therefore, this method is limited to detecting massive exoplanets on wide orbits. The measured intrinsic luminosity and large semi-major axis of directly imaged exoplanets are important quantities to understand their formation mechanism (Mordasini et al. 2015). The measured exoplanet luminosities and temperatures, combined with age-

5 Figure 1.2 An L band image of four directly-imaged giant exoplanets in the HR 8799 system from Marois et al. (2010). The planets are all in wide orbits around their host star, with projected separations between 24–68 au. A measure of their luminosities, in combination with age-dependent evolutionary models, enables estimates of planetary masses. dependent evolutionary models, enables estimates of planetary masses. However, these mass estimates are not direct measurements and are model-dependent, relying on theoretical models based on uncertain ages with large errors, resulting in considerable uncertainty in mass estimates for a substellar companion.

A total of 67 confirmed exoplanets have been discovered using this method, as indicated in the NASA Exoplanet Archive1. An example of directly imaged giant exoplanets on wide orbits is shown in Figure 1.2. The flux of the host star, HR 8799, was removed using PSF subtraction (Marois et al. 2010). The planets (labelled b, c, d, and e) are clearly visible in the image.

1.2.2 Transit Method

The transit method is a common technique that detects planets with nearly edge-on orbits.

This method relies on photometry to detect brief decreases in stellar flux as the planet passes in front of the host star. An example of a transit detection is shown in Figure 1.3.

1https://exoplanetarchive.ipac.caltech.edu

6 Figure 1.3 An example of a transit detection of a young sub-Neptune, K2-33b, from David et al. (2016). The exoplanet is detected from a decrease in brightness as it transits its host star, which allows for a measurement of its size.

The depth of the transit determines the size of the planet relative to that of the star:

∆F R 2 = p (1.1) F R? where ∆F/F is the ratio of the observed change in flux due to the transit, relative to the total stellar flux, Rp is the planet radius, and R? is the stellar radius. The time interval between consecutive transits determines the planet’s orbital period. This value, combined with Kepler’s third law, provides a measure of the orbital distance.

The transit method primarily detects giant planets in close-in orbits, since a short orbital period means the planet is more likely to transit during the observed time interval and a larger planet will create a more significant decrease in stellar flux during the transit. While this technique is complementary to the direct imaging method that finds giant planets on wide orbits, transit surveys are reliant on the geometry of the system (only exoplanets with orbits that are edge-on or nearly edge-on will be observed to transit their host star) and are also limited to relatively bright stars. The Kepler space telescope revolutionized transiting exoplanet science, finding the vast majority of the confirmed exoplanets detected to date.

The Transiting Exoplanet Survey Satellite, TESS, continued the search for transiting

7 exoplanets from space. While the re-purposed K2 mission and TESS have observed some

fields with young stars, most transit surveys are limited to mature stars that are well beyond

the planet formation evolutionary stage. Young exoplanet transit surveys are particularly

challenging because of the extreme variability of pre-main sequence stars (van Eyken et al.

2012).

1.2.3 Radial Velocity Method

The radial velocity (RV) method relies on spectroscopy to measure the line-of-sight

component of a star’s velocity. In a system with a planetary companion, the Doppler

shift of the spectral lines measures the star’s radial velocity as it orbits the common center

of mass of the star-planet system. An estimate of the companion mass can be derived from

the magnitude of the spectral line shift. A high-mass or close-in planet will induce a large

amplitude of RV variation. An example of RV variation indicative of a planet detection is

shown in Figure 1.4.

A planetary companion is indirectly detected by searching for repeating patterns in

the time-series RV measurements using periodogram analysis. This technique identifies a

periodic signal indicative of a companion’s orbital period. When the RVs are phased to this

period, the semi-amplitude of the RV variation and the stellar mass place a lower limit on

the companion mass, m2 sin i, where m2 is the companion mass and i is the inclination of the companion’s orbit relative to the plane of the sky. In addition to the orbital period, the semi-

major axis and the orbital eccentricity can also be identified from the RV measurements.

To characterize the companion mass and orbital properties, modulation in the RV phase

curve is represented by a Keplerian function:

2πG1/3 m sin i 1 K = 2 √ (1.2) P 2/3 1 − e2 m1 where K is the RV semi-amplitude derived from the RV curve, P is the companion orbital period determined from periodogram analysis of the RVs, m2 sin i is a limit on the

8 Figure 1.4 An example of a radial velocity detection of the young hot Jupiter, HD 285507 b, using optical data from Quinn et al. (2014). The RVs are phased to the companion period detected using periodogram analysis, and a fit to the curve is generated. The companion mass is measured from the amplitude of the RV curve.

9 companion mass, m1 is the star’s mass, and e is the orbital eccentricity of the companion. The semi-major axis may also be approximated using Kepler’s third law:

Gm  a3 = 1 P 2 (1.3) 4π2

The RV method is most sensitive to detecting companions with large masses and short orbital periods since increased mass and decreased distance results in a stronger gravitational effect of the companion on the host star’s motion. RV surveys, therefore, preferentially find high-mass, close-in companions. Similarly, the Doppler shift induced by a companion of a given mass and semi-major axis will be more significant for a low-mass star. While the RVs determine a lower limit on the planetary mass, m2 sin i, the true planetary mass is dependent on the inclination of the companion orbit. The companion

mass will be considerably larger than the lower limit if the orbit is nearly in the plane of

the sky, and will be close to the lower limit if the orbit is edge-on.

This exoplanet detection technique is complementary to the transit and direct detection

methods. Not many exoplanets are amenable to transit observations due to their geometries,

so the RV method can detect many planets that the transit method cannot. While the direct

detection technique is only able to find young planets in wide orbits, the RV technique favors

detections of planets in short orbits. Furthermore, while direct imaging relies on models to

determine a planetary mass, the RV method measures a model-independent limit on the

mass. To characterize the planet using the RV method, the stellar mass is needed. However,

this is generally much easier to measure than age, which is needed to characterize the mass

of directly imaged planets.

The RV method is particularly impactful for exoplanet characterization when combined

with transit observations. The combination of Doppler and transit measurements provides

the mass, radius, and therefore densities of a detected exoplanet, which can be used to infer

planetary compositions. Orbital properties and architectures, such as orbital separations,

eccentricities, and geometries can also be determined. By modeling the shape of the

10 observed transit light curve, the inclination of the orbit can be measured and combined with m2 sin i to calculate the true mass of the planetary companion. The survey presented in this dissertation relies on the RV method to detect young substellar companions. Since the RV method primarily identifies massive exoplanets on short-period orbits, these objects are likely to have migrated (Boss 1995), therefore a survey using this method can provide insights into the migration process and timescale. It is also a complementary approach to the direct detection technique, which has discovered several wide-orbit gas giant exoplanets. However, RV surveys require a large number of observations and a long enough observing baseline to sample many periods of the planetary companion.

Despite the advantages of studying young exoplanets to inform planet formation theory, the topic has remained relatively unexplored. There are several challenges to studying planets around young stars, many of which are still surrounded by a circumstellar disk, which accretes material onto the central star. These disks cause continuum veiling in spectra, infrared excess, and emission. Significant veiling can weaken the spectral lines while infrared excess and emission can contribute additional features to the spectrum.

Because young stars have stronger magnetic fields than mature stars, they consequently have higher levels of stellar activity. Young stars, therefore, display significant spectroscopic and photometric variability that can confuse the planet detection signals. They are also found at large distances from us (&120 pc), making them faint targets for observations. Infrared spectroscopy circumvents several of these challenges, particularly those related to stellar activity such as starspots (see Section 2.2 for additional discussion). Recent instrument developments resulting in high resolution infrared spectrographs have now made it feasible to perform a RV survey that mitigates some of the challenges of discovering young exoplanets around pre-main sequence stars.

11 1.3 Substellar Companions Around Young Stars

Despite the many hot/warm Jupiters discovered to date, the formation and evolution of these planets have been particularly challenging for theorists to explain. Observing and characterizing newly-formed planets around young stars is an important step in developing planet formation theory. Current theories, however, are primarily based on planetary systems that are billions of years old. Models must attempt to reproduce older systems with uncertain early conditions. It is therefore unclear which observed features are indicators of formation conditions, or of later orbital and physical evolution. The study of recently formed planets (.10 Myr), or the limits on occurrence rates based on non-detections, can provide insights into formation and evolution timescales and processes. Population statistics of young planetary companions can further inform these evolutionary processes and place constraints on formation mechanisms. While several young exoplanet candidates have been identified thus far, only about half a dozen young detections have been made using RV surveys. Given that an estimated 65% of Sun-like stars host a planet of any mass with P <

100 days (Mayor et al. 2011), observing pre-main sequence stars that will evolve to become solar analogues is a promising approach to finding young substellar companions.

1.3.1 Formation and Evolution Theory

Gas Giant Planet Formation

A formation mechanism for gas giant planets must explain how they can form on a relatively short timescale, while there is a sufficient disk to provide material for a planet’s high mass core and gaseous envelope. Most of the gas within the protoplanetary disk eventually disperses or accretes onto the host star within 10 Myr, so giant planets must form during this phase (Strom et al. 1993; Haisch et al. 2001). Observational evidence supports this timescale. Kruijer et al. (2017) found that Jupiter’s core likely formed rapidly within 1 Myr, according to meteoritic evidence. Observations of protoplanetary disks show evidence of

12 structure potentially caused by migrating planets, indicative of very early planet formation on a timescale of 0.5-2 Myr (Clarke et al. 2018; Fedele et al. 2018; Konishi et al. 2018).

A leading model for the formation of gas giant planets is core accretion (Mizuno et al.

1980; Pollack et al. 1996). This model predicts that gas giant planets form in a multi-stage process. This process starts with collisional growth of dust grains and proceeds through kilometer-sized planetesimals that form the planet core. Once the core is massive enough

(∼10 M⊕), it rapidly accretes surrounding gas from the protoplanetary disk to become a gas giant planet. An alternate model of gas giant planet formation is gravitational instability

(Boss 1997). In this scenario, instabilities in the rotating protoplanetary disk can lead to fragmentation. Eventually these fragments collapse and contract to form gas-accreting planetesimals that evolve to gas giant planets. This process is assumed to take place shortly after formation of the system when the protostellar disk mass is comparable to the mass of the central protostar, and therefore has dynamically significant self-gravity.

The core accretion and gravitational instability models of planetary formation are distinct in many ways. One notable difference is the timescale over which each are predicted to occur. The core accretion method is estimated to take several million years to form a gas giant planet (Dodson-Robinson et al. 2008). Specifically, at wide separations (>10 au) the timescale for forming gas giant planet cores is ∼10 Myr (Pollack et al. 1996; Alibert et al.

2005), which is older than the typical disk dissipation timescale of ∼0.1-10 Myr (Strom et al.

1993). Another 1000 years is required for gas accretion onto the core, and 100,000 years is needed for the eventual removal of the inner disk (Takeuchi et al. 1996), which creates a spatial separation between the planet and host star. This result suggests that gas giant planets are more likely to form after the protoplanetary disk has dispersed. However, several studies aiming to increase the efficiency of the core accretion method have successfully shown that this process can theoretically form Jupiter-mass planets at short orbital separations in disks with average lifetimes of ∼3 Myr (Alibert et al. 2005; Lissauer et al. 2009; Hasegawa

& Pudritz 2012), and could even form them on shorter timescales if given specific conditions in the system (Rice & Armitage 2003; Alibert et al. 2004; Dodson-Robinson et al. 2009).

13 Gravitational instability, on the other hand, predicts very fast formation times for massive planets in long-period orbits (103 years (Mayer et al. 2004) to 105 years (Bodenheimer

2006)), which suggests that gas giant planets are more likely to form early while the disk

still retains most of its initial mass.

Core accretion is generally favored as the dominant formation mechanism of gas giant

planets within a few au (Mordasini et al. 2012). This process is efficiently able to form high

mass cores, particularly in systems with high metallicity (Sato et al. 2005; Mordasini et al.

2012) because the metallicity of the disk shortens the solid accretion timescale during the

epoch of planet formation. This correlation with metallicity matches an observed planet-

metallicity correlation, in which gas giants have a higher frequency around metal-rich stars

(Fischer & Valenti 2005; Valenti & Fischer 2008; Wang & Fischer 2015). Alternatively,

gravitational instability is generally favored as the dominant mechanism for formation of

gas giants in wide orbits (Dodson-Robinson et al. 2009), such as those detected by direct

imaging. However, with a sufficiently massive disk it has been shown that this method may

form giant planets inside 20 au (Boss 2017, 2019). Since gravitational instability is expected

to occur on timescales that are orders of magnitude shorter than core accretion, this method

may be the favored theory to explain young gas giant detections that are only 1-2 Myr old

(Luhman et al. 2006; Baraffe et al. 2010; van Eyken et al. 2012). There is also evidence of

a mass dependence, with higher mass planets (above ∼4-10 MJ) forming by gravitational instability, and lower mass planets forming by core accretion (Schlaufman 2018). It is likely

that both mechanisms operate or dominate within different systems depending on their

early conditions.

Hot/Warm Jupiter Formation Mechanisms

The initial discoveries of close-in gas giant exoplanets were surprising in the context of

current planet formation theory. It is unlikely that a giant planet can form close to its parent

star for various reasons (Bodenheimer et al. 2000; Rafikov 2005, 2006): 1) temperatures near

the star are too high for the solid particles necessary for planet core formation to condense,

14 2) there is insufficient disk mass in the inner region to form a planet, 3) tidal interactions between the planet and star should cause the planet to migrate inward and be destroyed by the host star (Goldreich & Tremaine 1980; Ward 1997). Instead, it is widely thought that gas giant planets with orbital periods much less than 100 days form beyond the ice line at several au, where there are large quantities of rock, ice, and gases that can readily accumulate to form the planet core and envelope. The planet is then assumed to migrate inward to a short-period orbit after, or during, formation (Boss 1995).

A leading model to explain the existence of hot Jupiters is disk migration (Goldreich &

Tremaine 1980; Lin & Papaloizou 1986; Takeuchi et al. 1996; Ward 1997; Bodenheimer et al. 2000; Tanaka et al. 2002). In this scenario, the gas giant forms beyond 5 au and migrates inward through dynamical interactions with the protoplanetary disk until it reaches a stable orbit around its host star. The gravitational interactions between the disk and planet allow for an angular momentum exchange that moves the planet inward to a close-in orbit. Essentially, this process decreases the semi-major axis of the planet and also efficiently damps inclinations and eccentricities. As the planet increases in mass, it continues to migrate in a process known as Type II migration, in which the planet becomes massive enough to gravitationally open a gap in the disk, causing a barrier to flowing disk material (Ward 1997). To adequately explain the observed hot Jupiter population, a stopping mechanism to halt inward migration is needed. This stopping mechanism is not well understood, however, the gap created during Type II migration may serve this function

(Lin & Papaloizou 1986). Magnetic fields of pre-main sequence stars are typically strong enough to truncate the inner disk (Takeuchi et al. 1996), which may also be photoevaporated by irradiation from the host star (Matsuyama et al. 2003). This evacuated inner disk region may also stop migrating planets. Additionally, it is possible that many planets will migrate into the host star (Nelson et al. 2000). Disk migration must take place prior to the dissipation of the disk, on a timescale as short as 105 yrs (Ward 1997). It is predicted to occur immediately after or during the formation of a giant planet (Lufkin et al. 2004), and should form hot Jupiters with a low eccentricity and inclination (Lin et al.

15 1996). Observations, however, require an alternate migration scenario to explain the highly eccentric orbits (Naoz et al. 2011) and large obliquities (misalignments of the stellar spin axis and the planet orbital axis; Albrecht et al. 2012) of some observed hot Jupiters.

High-eccentricity migration has been proposed to explain the broad range of observed eccentricities in the hot Jupiter population. In this scenario, a giant planet forms at a large semi-major axis, is excited by secular processes to an inclined, highly eccentric orbit that shrinks and circularizes due to tidal dissipation caused by the host star. There are two main mechanisms that are theorized to increase the eccentricity of a gas giant planet: 1) planet-planet scattering between two or more planets, in which gravitational interactions between planets in a system causes chaotic orbital evolution that leads to planet scattering

(Rasio & Ford 1996; Wu & Murray 2003), or 2) the Kozai-Lidov effect: three-body secular interactions between the star, giant planet, and an inclined exterior companion, which produces oscillations resulting in periods of extreme orbital eccentricity (Wu & Murray 2003;

Fabrycky & Tremaine 2007). The inclined companion can be a stellar binary companion

(Naoz et al. 2012) or another planet (Naoz et al. 2011). High-eccentricity migration could also be the result of a combination of planet-planet scattering and the Kozai-Lidov effect

(Nagasawa et al. 2008). The eccentricity excitation and subsequent circularization occurs over a significantly longer timescale than disk migration: ∼100 Myr to >1 Gyr, depending

on the orbital and physical properties of the giant planet and the perturber (Fabrycky &

Tremaine 2007; Nagasawa et al. 2008). Likewise, migration of a giant planet may be the

result of a combination of disk migration and scattering by a stellar or planetary companion.

It is also possible that there are at least two populations of hot Jupiters that form by either

disk migration or high-eccentricity migration.

While migration mechanisms may explain how hot Jupiters reach their final orbits, they

can also cause the destruction of the planet. If a planet migrates to a position too close to

the host star, it can either be consumed by the central star, destroyed by tidal forces, ejected

out of the system, or it can experience extreme mass-loss from Roche lobe overflow (Holman

et al. 1997; Trilling et al. 1998; Kaib et al. 2013; Zuckerman 2014; Antonini et al. 2016;

16 Jackson et al. 2017). Close-in planets are subject to significant irradiation and tidal forces which can drive mass-loss. In addition to shortening the disk lifetime (Gorti et al. 2009), stellar irradiation can disrupt the gaseous envelopes of giant planets, resulting in partial removal of the planet’s atmosphere. While this effect may not be enough to significantly alter the planet evolution (hot Jupiters are predicted to lose only a few percent of their atmosphere on timescales on the order of a Gyr (Murray-Clay et al. 2009)), heating and tidal disruption of a hot Jupiter from the host star in a process known as tidal downsizing, can cause a hot Jupiter of 1 MJ to lose most of its gaseous envelope (Nayakshin 2011, 2017). WASP-12b is an example of an observed hot Jupiter undergoing orbital decay caused by tidal dissipation (Bailey & Goodman 2019).

As an alternative theory to migration, in situ formation has also been suggested as a possible formation mechanism for hot Jupiters. As described above, there are numerous challenges with forming a giant planet in a close-in orbit. However, the assumption that giant planet cores cannot form in the inner region of a stellar system is based on models that may not reflect a general property of all exoplanet systems (Chiang & Laughlin 2013).

Other processes have been proposed to explain how in situ formation of hot Jupiters can occur. Batygin et al. (2016) present a scenario in which a super-Earth sized core forms beyond the snow-line and simultaneously migrates inward with the disk material to a short- period orbit. It then accretes gas to become a hot Jupiter. Super-Earths are commonly found in close-in orbits (Fressin et al. 2013), which lends support to the possibility that they may serve as hot Jupiter cores in some systems. An alternate scenario also relies on ubiquitous observations of close-in lower mass exoplanets. Boley et al. (2016) propose that multiple low-mass planets in tightly-packed, short-period orbits can become unstable and consolidate within the gaseous disk to form hot Jupiter cores that then accrete a substantial gas envelope. This theory is further supported by evidence that both tightly-packed inner systems of planets and hot Jupiters are more frequent in metal-rich systems, and that high metallicity may increase the chance of collisions between small planets (Boley et al. 2016).

The timescale for in situ formation is ∼1-3 Myr (Bodenheimer et al. 2000; Batygin et al.

17 2016; Boley et al. 2016), which is considerably shorter than typical disk lifetimes. Therefore, in situ formation can produce hot Jupiters on relatively short timescales.

Some observational evidence exists in support of in situ formation. The accretion of pebbles has been shown to be a highly-efficient process, and may therefore be able to form hot Jupiter cores in the inner system (Lambrechts & Johansen 2012). Observed planetary systems, such as WASP-47, which is composed of multiple tightly-packed, massive planets on stable orbits, including a hot Jupiter, likely could not exist if formed by a disruptive migration mechanism (Becker et al. 2015; Sinukoff et al. 2017). Yet despite observational evidence supporting the possibility of in situ formation, unresolved problems still exist.

It is unclear what would stop a hot Jupiter that formed at such a short distance, from spiralling into its host star (Bodenheimer et al. 2000). This method also does not explain observed features of close-in giant planet orbits, such as the range of eccentricities and stellar spin-orbit misalignments that can readily be explained by high-eccentricity migration mechanisms (Winn et al. 2010).

Warm Jupiters have a larger orbital separation than hot Jupiters, and therefore may form and evolve through different mechanisms. A key difference in observations is that warm Jupiters have a large range of eccentricities, while hot Jupiters primarily have low eccentricities (Barker 2017). This effect is shown in Figure 1.5, taken from the Exoplanet

Orbit Database2. The hot Jupiters (P<10 days) favor orbits that are closer to circular than those found for the warm Jupiters (P>10 days).

It has been suggested that hot and warm Jupiters have a similar origin caused by high- eccentricity migration coupled with tidal friction, which circularizes the planetary orbit.

In this scenario, hot Jupiter orbits have been fully circularized, while warm Jupiters are still experiencing eccentricity oscillations induced by a perturber as they continue to evolve towards the hot Jupiter regime (Dawson & Chiang 2014; Dong et al. 2014). There has also been speculation that warm Jupiters may eventually evolve to become hot Jupiters

(Petrovich & Tremaine 2016; Masuda 2017), however, observed differences in hot/warm

2http://exoplanets.org

18 Figure 1.5 A comparison of the orbital eccentricity as a function of orbital period for hot (P<10 days) and warm Jupiters (P>10 days). The hot Jupiters tend to have lower eccentricities than the warm Jupiters, indicating a possible difference in their formation processes.

19 Jupiter properties suggests that they may in fact have two distinct formation pathways

(Antonini et al. 2016).

A leading theory for the formation of warm Jupiters is high-eccentricity migration. This mechanism is generally favored because of the significantly higher eccentricities measured for warm Jupiters (∼0.2, with some are as high as 0.8; Wright et al. 2012; Dawson &

Murray-Clay 2013; Anderson & Lai 2017). The detection of a higher rate of external giant planet companions also supports this scenario (Bryan et al. 2016; Huang et al. 2016), with either planet-planet scattering or the Kozai-Lidov mechanism increasing the eccentricity.

There are some problems with high-eccentricity migration as the formation mechanism of warm Jupiters. Antonini et al. (2016) found that most observed warm Jupiters with exterior planetary companions would not be stable if they had undergone high-eccentricity migration from beyond 1 au. Instead, warm Jupiters may form in situ or by disk migration, but with an exterior companion perturber that increases the eccentricity of the warm Jupiter’s orbit

(Anderson & Lai 2017). Huang et al. (2016) found that half the warm Jupiters in their sample have small companions, while nearly none of the hot Jupiters do, and therefore proposed that warm Jupiters may have formed in situ, which is less likely to disrupt the orbits of planetary companions. It has even been proposed that eccentric warm Jupiters form by a distinct method from circular warm Jupiters (Dawson & Murray-Clay 2013;

Petrovich & Tremaine 2016).

The dominant formation mechanism for hot/warm Jupiters has not yet been identified and is still heavily debated. Figure 1.6 illustrates possible formation scenarios for hot/warm

Jupiters, assuming the planet initially forms by core accretion. The diagram shows planet mass as a function of orbital period and indicates the possible formation mechanisms (either migration from beyond the snow line or in situ formation). While this figure illustrates the broad consensus on giant planet formation, the exact mechanisms of formation are still unknown.

Disk migration is supported by observational results, such as hot Jupiters that exist in co-planar orbits, however, a wide distribution of alignments with the observed host stars

20 Figure 1.6 An illustration of favored formation scenarios for hot and warm Jupiters from Wu et al. (2018). This diagram shows planet mass as a function of orbital period with the hot Jupiters in the shaded orange region. These planets likely formed beyond the iceline by the core accretion method, then migrated inward either by disk migration or high-eccentricity migration. In situ formation is also a possibility. Warm Jupiters (and other exoplanets within 10-100 day orbital periods), discovered by Wu et al. (2018), are indicated as yellow- filled red circles. Warm Jupiters may have a distinct formation mechanism than the hot Jupiters.

21 suggests a more turbulent migration mechanism (Johnson et al. 2009; Triaud et al. 2010).

Misalignments between the planet orbit and stellar rotation axis, as well as the high orbital eccentricities that are sometimes observed, provide evidence for high-eccentricity migration.

Yet evidence against high-eccentricity migration has also been found. Simulations show that this mechanism can only account for <10% of all gas giants observed between 0.1 and 1

au, and more likely results in planet ejections and collisions with the host star rather than

tidal migration (Antonini et al. 2016). It has also been speculated that disk migration

may instead lead to the observed misalignments, which may be caused by tidal dissipation

from turbulence in the disk (McKee & Ostriker 2007; Fielding et al. 2015). The lack of

detected companions to hot/warm Jupiters also suggests another migration mechanism for

the majority of systems (Knutson et al. 2014; Ngo et al. 2016). Alternatively, the lack of

companions could indicate that the single highly-eccentric giant planet is the sole survivor

of a scattering event that ejected the other planet out of the system (Howard 2013). While

several channels of formation may produce hot and warm Jupiters, it is difficult to discern

which mechanism is dominant, and even which is responsible for any particular system.

A key difference between disk migration and high-eccentricity migration is the timescale

over which they occur (<10 Myr (Ward 1997) and ∼100 Myr to over 1 Gyr (Fabrycky

& Tremaine 2007; Nagasawa et al. 2008), respectively). Young exoplanet detections are

crucial for identifying formation mechanisms based on timescales constrained by the age

of the system. RV surveys provide mass and orbital properties, such as eccentricity, which

can further inform evolution theory and potentially rule out certain formation scenarios.

This approach offers a more immediate view of formation rather than attempting to discern

unknown initial conditions from mature systems that have already undergone substantial

evolution.

Brown Dwarf Formation and the Origin of the Brown Dwarf Desert

The presence of the brown dwarf desert (the region within .3 au of solar-type stars in which there is an observed scarcity of brown dwarfs) supports the conventional belief that brown

22 dwarfs and gas giant planets have distinct formation mechanisms, and also suggests that some aspect of the formation or early evolution of brown dwarfs differs from that of stars; it is possible to explain the brown dwarf desert by postulating that the formation mechanism for brown dwarfs is completely different from that of both stars and giant planets (Armitage

& Bonnell 2002). Ma & Ge (2014) suggest that brown dwarfs with masses below 42.5 MJ form by a process similar to some gas giant planets (disk instability), while more massive brown dwarfs form like low-mass stars (cloud fragmentation), based on observed depletion between these two mass regimes. This result, however, has been called into question based on their small sample size and statistical methods (Carmichael et al. 2019). Schlaufman

(2018) determined that planetary objects with mass .4 MJ likely formed by core accretion, as suggested by the observed correlation with host star metallicities, while objects with mass &10 MJ likely formed instead by gravitational instability.

A better understanding of brown dwarf formation can lead to more informed definitions to distinguish between different objects. Currently, the standard definition of a brown dwarf relies on mass limits, however, mass boundaries do not necessarily correspond to transitions in the mode of formation, and the physics of formation mechanisms is responsible for the relative abundances of objects of varying masses rather than fusion onset limits

(Grether & Lineweaver 2006). Understanding the origin of the brown dwarf desert may provide information on brown dwarf formation, and further distinguish characteristics between brown dwarfs, planets, and stars, besides their mass and ability to fuse deuterium

(Carmichael et al. 2019).

Detections of young, close-in brown dwarfs may also provide clues as to the origin of the brown dwarf desert. It has been suggested that the brown dwarf desert is a reflection of the boundary between the high-mass end of the giant planet distribution, and the low-mass end of the stellar distribution (Persson et al. 2019). However, it is also possible that the brown dwarf desert was created by some later evolutionary mechanism. Brown dwarf companions are thought to form at the same time as the host star, followed by inward migration. Because the brown dwarf likely formed simultaneously with the central star, rather than at a later

23 epoch, the disk dispersal and stopping mechanisms that allow hot Jupiters to survive have not yet occurred in the system. Therefore, the brown dwarf is more likely to migrate until it has merged with the host star (Armitage & Bonnell 2002). Similarly, dynamical evolution between a star-brown dwarf system and other stars within a newly-formed cluster may eventually eject the brown dwarf from the system; those companions that aren’t ejected may instead continue to accrete until they eventually reach stellar masses (Reipurth & Clarke

2001). These scenarios describe a theory for the origin of the brown dwarf desert and can be tested by searching for close-in brown dwarf companions in young systems. Armitage &

Bonnell (2002) predict that the timescale for destruction of close-in brown dwarfs is 1 Myr.

As a result, pre-main sequence stars should have an order of magnitude more brown dwarfs in close orbits than their main sequence counterparts. By considering the shape of the brown dwarf desert in mass-period parameter space, Shahaf & Mazeh (2019) could attribute the observed boundaries to the effectiveness of the destructive mechanism weakening at longer distances, while initial conditions did not easily explain those boundaries. This supports the scenario that later evolution forms the brown dwarf desert, suggesting that the close-in brown dwarf occurrence rate may be higher in pre-main sequence systems.

1.3.2 Occurrence Rates of Substellar Companions

Occurrence rates of planetary companions can serve as a test of evolution and formation theory. Population synthesis studies provide a basis to test simulations of evolutionary processes, by requiring that the efficiency of forming hot Jupiters in model simulations matches the observed occurrence rate (Coleman & Nelson 2016). In particular, the study of young planetary populations may reveal a difference in the occurrence rate of substellar companions in pre-main sequence and main sequence stars. If the distribution of observed planets around main sequence stars is shaped by processes that take place after the circumstellar disk disperses (such as high-eccentricity migration), then the occurrence rate of planets around pre-main sequence stars that still have, or recently lost, their circumstellar disks will be different. Furthermore, the timescales of different formation mechanisms for

24 hot Jupiters vary. In the case of young systems, some evolutionary processes take longer than the age of the system. Therefore, finding a higher occurrence rate of hot Jupiters in young systems would favor a formation mechanism that happens on a short timescale

(within the age of the system). In the case of a young hot Jupiter discovered around a

<5 Myr old star, disk migration would be favored over high-eccentricity migration (in situ migration is also a possibility, but this method is not generally favored because of the challenges in forming gas giant planets close to their host stars).

Hot/warm Jupiters and close-in brown dwarfs may be more prevalent around pre- main sequence stars, as we are more likely to observe them before they are destroyed or transformed by evolutionary processes. A difference in occurrence rate in young systems can place constraints on the efficiency and timescale of a migration mechanism, or the propensity of a disk to form giant planets given certain initial conditions. A different occurrence rate based on age can be caused by inward migration resulting in a merger with the host star (Trilling et al. 1998, Armitage & Bonnell 2002, Heller 2019), or extreme mass- loss by photoevaporation causing the hot Jupiter to evolve to a lower-mass super-Earth

(Nayakshin 2011).

A hot Jupiter occurrence rate of 1.2 ± 0.4% for Sun-like stars in the solar neighborhood has been determined from a ground-based radial velocity survey (Wright et al. 2012), consistent with other results (1.2%, Marcy et al. 2005; 1.5%, Cumming et al. 2008; 0.9%,

Mayor et al. 2011). A survey of transiting planets from the Kepler and TESS space missions found a hot Jupiter occurrence rate which is about half that estimated by RV surveys

(0.4 ± 0.1%; Howard et al. 2012; Zhou et al. 2019). This rate was confirmed by a ground- based transiting survey (0.3%; Gould et al. 2006).

A difference in hot Jupiter occurrence rates estimated using different detection methods may reflect limitations in the survey sample sizes; another potential cause is a difference in the stellar populations or parameters of these surveys, particularly the metallicity (Wright et al. 2012). An increased frequency of giant planets as a function of the stellar metallicity,

[Fe/H], has been established for solar-type stars, with a rate as high as 30% for stars

25 with 2x solar metallicity compared to ∼3% for stars with solar metallicity (Santos et al.

2004; Fischer & Valenti 2005; Johnson et al. 2010). A correlation between the metallicity

and occurrence of giant planets suggests the core accretion method of formation since this

process is expected to be more efficient in high-metallicity environments (Mayor et al.

2011). Gould et al. (2006) estimated a significant bias toward higher metallicity stars in

RV surveys based on target choice and stellar population models of the Milky Way. This

bias could explain why the frequency of hot Jupiters is higher in RV surveys and instead,

this result may actually reflect the higher occurrence rate of hot Jupiters around metal-rich

stars. However, Guo et al. (2017) measured the metallicities of Kepler stars and found that the difference between the mean metallicity of the California Planet Search RV survey and the mean Kepler metallicities was insufficient to explain the occurrence rate disparity.

Instead, they suggest the difference in occurrence rates may be a combination of other factors, such as higher stellar multiplicity in the Kepler sample leading to suppressed hot

Jupiter occurrence, imprecise determination of stellar properties, or a difference in mass distributions between the transiting sample and the RV sample.

Population studies of pre-main sequence stars have produced some promising results thus far. Observational evidence suggests that young hot Jupiter occurrence rates may be higher than those measured in mature systems. Recent young planet discoveries, in comparison to sample sizes, suggests that hot Jupiters may be more frequent around young stars with ages ∼2-20 Myr (Donati et al. 2017; Yu et al. 2017). Quinn et al. (2014) estimate a hot

Jupiter frequency of ∼2% in the metal-rich open clusters, Praesepe and Hyades (with ages of several hundred Myr), however, when adjusted for the high metallicity of these regions, the frequency is closer to that of mature field stars (∼1%). Similarly, Grandjean et al. (2020) measured a giant planet and brown dwarf occurrence rate (also for stars with ages of several hundred Myr) that is consistent with the main sequence rate. Simulations of the inner disk and combined torques on a planet during disk migration predict a hot Jupiter formation occurrence rate between 8-41%, with a 3-15% chance of survival during disk migration

(Heller 2019). In systems where tidal torques are too weak to stop disk migration, Heller

26 (2019) determined a destruction timescale of ∼10 Myr. Therefore, a survey of stars younger than these ages may result in an increased occurrence rate. A large-scale survey of pre-main sequence stars is needed to determine if a difference in hot Jupiter occurrence rates based on age exists and why.

Warm Jupiters seem to significantly outnumber hot Jupiters, at least in main-sequence systems (Han et al. 2014; Santerne et al. 2016; Anderson & Lai 2017). The ratio of warm

Jupiters to hot Jupiters (WJ/HJ) varies based on the exact constraints in defining these two populations. Han et al. (2014) estimated the WJ/HJ ratio to be ∼3.9 for warm Jupiters

with a mass >0.5 MJ. Adopting a more conservative definition of a warm Jupiter (0.1 au < a < 0.5 au), Anderson & Lai (2017) found a WJ/HJ ratio of ∼1.6. This ratio may be

higher if selection effects are accounted for (Anderson & Lai 2017). Santerne et al. (2016)

estimated a WJ/HJ ratio of ∼8.3, defining hot Jupiters as gas giant planets with periods

<10 days and warm Jupiters as gas giant planets with larger periods. These ratio estimates

can be the result of differences or similarities in hot and warm Jupiter formation, and can

also be related to the early conditions in planet-forming systems (Ali-Dib et al. 2017). An

example of the occurrence rates of hot Jupiters and warm Jupiters as a function of orbital

period from Santerne et al. (2016) is shown in Figure 1.7. The number of gas giant planets

drops at the boundary between hot and warm Jupiters, and then rapidly increases at longer

orbital separations.

The population statistics of close-in brown dwarfs around main sequence stars has

yielded surprising results. These studies have identified the region with orbital separation

<3 au as the brown dwarf desert. While ∼16% of solar-type stars have close companions

with an orbital period <5 years, less than 1% of those companions are brown dwarfs; ∼11%

are stellar and ∼5% are giant planets (Grether & Lineweaver 2006). Grieves et al. (2017)

estimated that the brown dwarf desert occurrence rate around solar-type stars with periods

.300 days is only 0.56%.

The paucity of brown dwarfs at small orbital distances is not predicted by formation

theory. Theory predicts that if brown dwarfs follow the same distribution of separations as

27 Figure 1.7 An example of the occurrence rates of hot Jupiters (P<10 days) and warm Jupiters (P>10 days) (as determined by Santerne et al. (2016)) as a function of orbital period. This figure was reproduced by Ali-Dib et al. (2017) who overplotted planet formation models to test initial conditions of planet-forming material. Population statistics allowed them to show that planet-forming seeds in the disk likely preferentially group at the H20 (blue) and CO (green) icelines rather than linearly throughout the disk (red).

28 stars, the number of brown dwarfs in close orbits ought to be roughly an order of magnitude higher (Gizis et al. 2001; Armitage & Bonnell 2002). It has therefore been suggested that there is an evolutionary cause of the brown dwarf desert that depletes the population at later ages. As noted above, this implies that close-in brown dwarfs may be up to an order of magnitude more common around pre-main sequence stars. These brown dwarfs may later become depleted as a consequence of inward migration that leads to a merger with the host star. Gas giant planets may escape this fate by forming at later epochs when the circumstellar disk begins to disperse and form lower mass objects (Armitage & Bonnell

2002).

1.3.3 Young Substellar Candidate Discoveries

Several young substellar companions in wide orbits have been detected using imaging techniques. Chauvin et al. (2004, 2005) directly imaged a distant giant planet candidate comoving with an 8 Myr old brown dwarf in the young TW Hya association. Young brown dwarf companions in wide orbits have also been detected, such as the ∼4 Myr old 30-50

MJ DH Tau b at an orbital separation of 330 au (Itoh et al. 2005) and the ∼2 Myr old substellar companion to CHXR 73 (Luhman et al. 2006). Liu et al. (2013) discovered a

young free-floating brown dwarf with no host star, PSO J318.5-22, with a mass of ∼7 MJ and an age between 8-20 Myr. The LkCa 15 system has been extensively studied to detect

and characterize protoplanetary companions. Kraus & Ireland (2012) discovered a young

exoplanet candidate inside a known gap within the system’s protoplanetary disk with an

estimated mass of 6 MJ and an age of only 1 Myr. This planet is still in its early formation phase and could be accreting circumstellar material (Kraus & Ireland 2012; Ireland &

Kraus 2014; Sallum et al. 2015). Sallum et al. (2015) suggest that there could be multiple

accreting protoplanets in this system, between 15 and 19 au. These are some of the youngest

detections discovered to date, but several other young exoplanets have also been imaged

in older systems, such as Fomalhaut b (∼60 Myr old; Kalas et al. 2005) and a series of

planets in the HR 8799 system (∼30-60 Myr old; Marois et al. 2008; Marois et al. 2010; see

29 Figure 1.2). While the direct imaging technique has produced many substellar companion detections, it is limited to objects in wide orbits around their host star.

Transit photometry has led to several young exoplanet candidate discoveries in close-in orbits, predominantly using data from the K2 Mission (Howell et al. 2014). Mann et al.

(2016a) discovered a Neptune-sized exoplanet, K2-25b, transiting an M dwarf in the Hyades open cluster (650-800 Myr old; Perryman et al. 1998) with an orbital period of 3.5 days.

This exoplanet has a larger radius than other planets with similar orbital periods, which suggests that close-in planets lose some of their atmosphere within the first few hundred million years (Mann et al. 2016a). A transit discovery of a 5-10 Myr old super-Neptune, K2-

33b, in a 5.4-day orbital period in the Upper Sco region was reported by Mann et al. (2016b) and David et al. (2016). This detection suggests that close-in planets can either form in situ or migrate within the age of the system (∼10 Myr), likely through disk migration since other

dynamical migration mechanisms take longer than the age of the planet. Since it is rare to

detect planets of this size close to mature low-mass stars (Dressing & Charbonneau 2015),

it has been suggested that K2-33b is still contracting and losing some of its atmosphere, or

that it is still undergoing radial migration (David et al. 2016; see Figure 1.3). Transiting

close-in giant planets have also been found around low-mass stars, such as the M-dwarf,

NGTS-1 (Bayliss et al. 2018). This suggests that hot Jupiters form and evolve in a similar

manner around low-mass stars as they do around solar-type stars. Recently, David et al.

(2019) reported the discovery of a warm Jupiter in the Taurus-Auriga region transiting the

pre-main sequence star, V1298 Tau, which is a solar-type star with an estimated age of 23

Myr.

Yet another detection of a transiting exoplanet detection has proven quite controversial.

A transit detection of a hot Jupiter, PTFO 8-8695, in a 0.5 day orbit around a 3 Myr old T

Tauri star in the Orion region (van Eyken et al. 2012; Barnes et al. 2013; Ciardi et al. 2015)

may present an example of mass loss and evaporation as the planet spirals into its host star

(Johns-Krull et al. 2016b). The light curves of the host star vary significantly from transit

to transit, which has been attributed to the companion transiting either various features on

30 the stellar photosphere (van Eyken et al. 2012) or the gravity-darkened stellar disk (Barnes et al. 2013; although also see Howarth 2016 for an alternate view). This variation may also be caused by a precession between the stellar rotation axis and planetary orbital plane

(Barnes et al. 2013; Ciardi et al. 2015). The detection of excess H-α emission that moves in velocity as expected for an exoplanet seemingly confirms the planetary nature of the object

(Johns-Krull et al. 2016b), however, the detection has been called into question and may instead be indicative of stellar activity features, eclipses by a circumstellar dust clump, or occultations of an accretion hotspot (Yu et al. 2015; Onitsuka et al. 2017). If real, this planet candidate would offer a view of an exoplanet in an early phase of destruction caused by its proximity to its host star. This detection illustrates the potential challenges of the transit method for young exoplanet discoveries.

Young transiting planets discovered thus far seem to have a larger radius than their more mature counterparts with similar orbital periods (Mann et al. 2016b). This result suggests that evolutionary factors, such as photoevaporation by the host star, may significantly impact the exoplanet atmosphere in the first few Gyr (Owen & Wu 2013). Evaporation rates are strongly dependent on the mass of the planet, so follow-up with the RV method is important to further test this scenario (Mann et al. 2016b), and is also important for candidate confirmation.

The RV method has also proven to be a promising technique for detecting young close-in exoplanets. Quinn et al. (2014) used optical spectroscopy to search for young companions in the 650–800 Myr old Hyades open cluster and reported a discovery of a hot Jupiter with a mass of 1 MJ and a 6-day orbital period, HD 285507 b (see Figure 1.4). This discovery suggests that a high density of stars does not inhibit the formation and migration of giant planets. Migration caused by interactions with a third body is the likely method of formation of this companion because of its high eccentricity and young age (there has not been sufficient time in this system for the orbit of the planet to circularize). Similarly,

Donati et al. (2016, 2017) used optical spectroscopy combined with Zeeman Doppler Imaging

(an imaging technique used to map stellar activity features on a star’s surface) to detect

31 a 0.8 MJ hot Jupiter candidate, V830 Tau b, around a 2 Myr old star. The young age of the system suggests that gas driven migration was the cause of the planet’s close orbit, and likely happened very rapidly after formation. As part of the same survey, and using the same techniques, Yu et al. (2017) discovered a 1.7 MJ planet candidate, Tap-26, in an 11-day orbit around a 17 Myr old star. There are, however, many challenges associated with interpreting RVs of young systems, and an independent RV survey was unable to confirm the V830 Tau b detection (Damasso et al. 2020).

While most RV surveys have relied on optical spectroscopy, infrared spectroscopy is still relatively unexplored. Johns-Krull et al. (2016a) discovered an ∼11 MJ young hot Jupiter candidate in a 9-day orbital period around CI Tau, a 2 Myr old star in the Taurus region.

Almeida et al. (2017) also discovered a ∼19 MJ substellar companion with a 25-day orbital period close to the inner circumstellar disk radius around AS205A, a young star in a triple system that is only ∼0.5 Myr old. Both discoveries place constraints on the timescale of planet and brown dwarf formation.

The region around solar-type stars known as the brown dwarf desert is expected to be depleted but not empty, and some brown dwarfs within this region have been detected. A substellar companion with mass ∼17 MJ was found at a distance of <3 au in the HD 168443

system (Marcy et al. 2001). Udry et al. (2002) detected a ∼17 MJ companion orbiting HD 202206 at a distance of <1 au. Another companion, EPIC 212036875b with a mass ∼52

MJ, was detected at a distance of 0.06 au from its host star, indicating that either this is a particularly rare object, or the brown dwarf desert is more populated than previously believed (Carmichael et al. 2019; Persson et al. 2019). Additionally, a large-scale RV survey detected a total of 10 brown dwarfs with orbital distances <1 au (Grieves et al. 2017). A preliminary survey of the brown dwarf desert in pre-main sequence systems did not result in any additional discoveries (Mahmud et al. 2011).

32 1.4 Science Objectives

As outlined in this chapter, a search for substellar companions around young stars using infrared spectroscopy has strong potential to contribute to formation theories of hot/warm

Jupiters and brown dwarfs. The broad science objectives for this dissertation are to:

1) Detect, confirm, and characterize young gas giants and brown dwarfs using the radial velocity method and place limits on possible companions not detected by this survey. The

field of exoplanets orbiting young stars is still in its early stages and only a small number of young hot Jupiters have been discovered to date. Any additional detections or limits on the occurrence of hot Jupiters, warm Jupiters, and close-in brown dwarfs will meaningfully contribute to the theory of planet/brown dwarf formation and evolution.

2) Characterize potential pre-main sequence host stars. The radial velocity method relies on spectroscopy of young stars to detect planet signatures. This spectroscopy is also useful for characterizing stellar properties of the potential host stars, such as the rotational velocity and effective temperature. Additionally, the stellar rotation period in combination with the rotational velocity can place a limit on the stellar radius.

3) Provide observational results that may be extended to planet formation theory, planet statistics studies, and the origin of the brown dwarf desert. Detections of young exoplanets can be used to test theories of evolutionary processes, such as disk migration, scattering, and other dynamical interactions. Observing evidence of these processes in early systems is important for understanding evolutionary factors and their timescales. While hot Jupiters around pre-main sequence stars are rare, young hot Jupiter discoveries to date suggest that the occurrence rate could be higher in pre-main sequence systems. Even if no further young planet detections are made, this work will place constraints on the occurrence rates for young stars. Furthermore, detections of young brown dwarfs in close-in orbits can inform the origin of the brown dwarf desert.

33 References

Adams, F. C., Ruden, S. P., & Shu, F. H. 1989, ApJ, 347, 959

Agnew, M. T., Maddison, S. T., Thilliez, E., & Horner, J. 2017, MNRAS, 471, 4494

Albrecht, S., Winn, J. N., Johnson, J. A., Howard, A. W., Marcy, G. W., Butler, R. P.,

Arriagada, P., Crane, J. D., Shectman, S. A., Thompson, I. B., Hirano, T., Bakos, G., &

Hartman, J. D. 2012, ApJ, 757, 18

Ali-Dib, M., Johansen, A., & Huang, C. X. 2017, MNRAS, 469, 5016

Alibert, Y., Mordasini, C., & Benz, W. 2004, A&A, 417, L25

Alibert, Y., Mordasini, C., Benz, W., & Winisdoerffer, C. 2005, A&A, 434, 343

Almeida, P. V., Gameiro, J. F., Petrov, P. P., Melo, C., Santos, N. C., Figueira, P., &

Alencar, S. H. P. 2017, A&A, 600, A84

Alvarado-Montes, J. A., Zuluaga, J. I., & Sucerquia, M. 2017, MNRAS, 471, 3019

Anderson, K. R. & Lai, D. 2017, MNRAS, 472, 3692

Antonini, F., Hamers, A. S., & Lithwick, Y. 2016, AJ, 152, 174

Armitage, P. J. & Bonnell, I. A. 2002, MNRAS, 330, L11

Bailey, A. & Goodman, J. 2019, MNRAS, 482, 1872

Baraffe, I., Chabrier, G., & Barman, T. 2010, Reports on Progress in Physics, 73, 016901

34 Barker, A. J. 2017, arXiv e-prints, arXiv:1703.08003

Barnes, J. W. & O’Brien, D. P. 2002, ApJ, 575, 1087

Barnes, J. W., van Eyken, J. C., Jackson, B. K., Ciardi, D. R., & Fortney, J. J. 2013, ApJ,

774, 53

Bate, M. R., Bonnell, I. A., & Bromm, V. 2003, MNRAS, 339, 577

Batygin, K., Bodenheimer, P. H., & Laughlin, G. P. 2016, ApJ, 829, 114

Bayliss, D., Gillen, E., Eigm¨uller,P., McCormac, J., Alexander, R. D., Armstrong, D. J.,

Booth, R. S., Bouchy, F., Burleigh, M. R., Cabrera, J., Casewell, S. L., Chaushev, A.,

Chazelas, B., Csizmadia, S., Erikson, A., Faedi, F., Foxell, E., G¨ansicke, B. T., Goad,

M. R., Grange, A., G¨unther, M. N., Hodgkin, S. T., Jackman, J., Jenkins, J. S., Lambert,

G., Louden, T., Metrailler, L., Moyano, M., Pollacco, D., Poppenhaeger, K., Queloz, D.,

Raddi, R., Rauer, H., Raynard, L., Smith, A. M. S., Soto, M., Thompson, A. P. G.,

Titz-Weider, R., Udry, S., Walker, S. R., Watson, C. A., West, R. G., & Wheatley, P. J.

2018, MNRAS, 475, 4467

Becker, J. C., Vanderburg, A., Adams, F. C., Rappaport, S. A., & Schwengeler, H. M. 2015,

ApJ, 812, L18

B´ejar,V. J. S., Mart´ın,E. L., Zapatero Osorio, M. R., Rebolo, R., Barrado y Navascu´es,

D., Bailer-Jones, C. A. L., Mundt, R., Baraffe, I., Chabrier, C., & Allard, F. 2001, ApJ,

556, 830

Bodenheimer, P. 2006, Historical notes on planet formation, ed. H. Klahr & W. Brandner,

1

Bodenheimer, P., Hubickyj, O., & Lissauer, J. J. 2000, , 143, 2

Boley, A. C., Granados Contreras, A. P., & Gladman, B. 2016, ApJ, 817, L17

Boss, A. P. 1995, Science, 267, 360

35 —. 1997, Science, 276, 1836

—. 2017, ApJ, 836, 53

—. 2019, ApJ, 884, 56

Bryan, M. L., Knutson, H. A., Howard, A. W., Ngo, H., Batygin, K., Crepp, J. R., Fulton,

B. J., Hinkley, S., Isaacson, H., Johnson, J. A., Marcy, G. W., & Wright, J. T. 2016,

ApJ, 821, 89

Burrows, A., Hubbard, W. B., Lunine, J. I., & Liebert, J. 2001, Reviews of Modern Physics,

73, 719

Carmichael, T. W., Latham, D. W., & Vand erburg, A. M. 2019, AJ, 158, 38

Chauvin, G., Lagrange, A.-M., Dumas, C., Zuckerman, B., Mouillet, D., Song, I., Beuzit,

J.-L., & Lowrance, P. 2004, A&A, 425, L29

—. 2005, A&A, 438, L25

Chiang, E. & Laughlin, G. 2013, MNRAS, 431, 3444

Ciardi, D. R., van Eyken, J. C., Barnes, J. W., Beichman, C. A., Carey, S. J., Crockett,

C. J., Eastman, J., Johns-Krull, C. M., Howell, S. B., Kane, S. R., . Mclane, J. N.,

Plavchan, P., Prato, L., Stauffer, J., van Belle, G. T., & von Braun, K. 2015, ApJ, 809,

42

Clarke, C. J., Tazzari, M., Juhasz, A., Rosotti, G., Booth, R., Facchini, S., Ilee, J. D.,

Johns-Krull, C. M., Kama, M., Meru, F., & Prato, L. 2018, ApJ, 866, L6

Coleman, G. A. L. & Nelson, R. P. 2016, MNRAS, 457, 2480

Cumming, A., Butler, R. P., Marcy, G. W., Vogt, S. S., Wright, J. T., & Fischer, D. A.

2008, PASP, 120, 531

36 Damasso, M., Lanza, A. F., Benatti, S., Rajpaul, V. M., Mallonn, M., Desidera, S.,

Biazzo, K., D’Orazi, V., Malavolta, L., Nardiello, D., Rainer, M., Borsa, F., Affer, L.,

Bignamini, A., Bonomo, A. S., Carleo, I., Claudi, R., Cosentino, R., Covino, E., Giacobbe,

P., Gratton, R., Harutyunyan, A., Knapic, C., Leto, G., Maggio, A., Maldonado, J.,

Mancini, L., Micela, G., Molinari, E., Nascimbeni, V., Pagano, I., Piotto, G., Poretti,

E., Scandariato, G., Sozzetti, A., Capuzzo Dolcetta, R., Di Mauro, M. P., Carosati, D.,

Fiorenzano, A., Frustagli, G., Pedani, M., Pinamonti, M., Stoev, H., & Turrini, D. 2020,

arXiv e-prints, arXiv:2008.09445

Darriba, L. A., de El´ıa,G. C., Guilera, O. M., & Brunini, A. 2017, A&A, 607, A63

David, T. J., Cody, A. M., Hedges, C. L., Mamajek, E. E., Hillenbrand, L. A., Ciardi,

D. R., Beichman, C. A., Petigura, E. A., Fulton, B. J., Isaacson, H. T., Howard, A. W.,

Gagn´e,J., Saunders, N. K., Rebull, L. M., Stauffer, J. R., Vasisht, G., & Hinkley, S.

2019, AJ, 158, 79

David, T. J., Hillenbrand, L. A., Petigura, E. A., Carpenter, J. M., Crossfield, I. J. M.,

Hinkley, S., Ciardi, D. R., Howard, A. W., Isaacson, H. T., Cody, A. M., Schlieder, J. E.,

Beichman, C. A., & Barenfeld, S. A. 2016, Nature, 534, 658

Dawson, R. I. & Chiang, E. 2014, Science, 346, 212

Dawson, R. I. & Murray-Clay, R. A. 2013, ApJ, 767, L24

Dodson-Robinson, S. E., Bodenheimer, P., Laughlin, G., Willacy, K., Turner, N. J., &

Beichman, C. A. 2008, ApJ, 688, L99

Dodson-Robinson, S. E., Veras, D., Ford, E. B., & Beichman, C. A. 2009, ApJ, 707, 79

Donati, J. F., Moutou, C., Malo, L., Baruteau, C., Yu, L., H´ebrard,E., Hussain, G.,

Alencar, S., M´enard,F., Bouvier, J., Petit, P., Takami, M., Doyon, R., & Cameron,

A. C. 2016, Nature, 534, 662

37 Donati, J.-F., Yu, L., Moutou, C., Cameron, A. C., Malo, L., Grankin, K., H´ebrard,E.,

Hussain, G. A. J., Vidotto, A. A., Alencar, S. H. P., Haywood, R. D., Bouvier, J., Petit,

P., Takami, M., Herczeg, G. J., Gregory, S. G., Jardine, M. M., & Morin, J. 2017,

MNRAS, 465, 3343

Dong, S., Katz, B., & Socrates, A. 2014, ApJ, 781, L5

Dressing, C. D. & Charbonneau, D. 2015, ApJ, 807, 45

Fabrycky, D. & Tremaine, S. 2007, ApJ, 669, 1298

Fedele, D., Tazzari, M., Booth, R., Testi, L., Clarke, C. J., Pascucci, I., Kospal, A., Semenov,

D., Bruderer, S., Henning, T., & Teague, R. 2018, A&A, 610, A24

Fielding, D. B., McKee, C. F., Socrates, A., Cunningham, A. J., & Klein, R. I. 2015,

MNRAS, 450, 3306

Fischer, D. A. & Valenti, J. 2005, ApJ, 622, 1102

Fogg, M. J. & Nelson, R. P. 2007, A&A, 461, 1195

Fressin, F., Torres, G., Charbonneau, D., Bryson, S. T., Christiansen, J., Dressing, C. D.,

Jenkins, J. M., Walkowicz, L. M., & Batalha, N. M. 2013, ApJ, 766, 81

Georgakarakos, N., Eggl, S., & Dobbs-Dixon, I. 2018, ApJ, 856, 155

Gizis, J. E., Kirkpatrick, J. D., Burgasser, A., Reid, I. N., Monet, D. G., Liebert, J., &

Wilson, J. C. 2001, ApJ, 551, L163

Goldreich, P. & Tremaine, S. 1980, ApJ, 241, 425

Gong, Y.-X., Zhou, J.-L., Xie, J.-W., & Wu, X.-M. 2013, ApJ, 769, L14

Gonzalez, G., Brownlee, D., & Ward, P. 2001, , 152, 185

Gorti, U., Dullemond, C. P., & Hollenbach, D. 2009, ApJ, 705, 1237

38 Gould, A., Dorsher, S., Gaudi, B. S., & Udalski, A. 2006, , 56, 1

Grandjean, A., Lagrange, A. M., Keppler, M., Meunier, N., Mignon, L., Borgniet, S.,

Chauvin, G., Desidera, S., Galland, F., Messina, S., Sterzik, M., Pantoja, B., Rodet, L.,

& Zicher, N. 2020, A&A, 633, A44

Grether, D. & Lineweaver, C. H. 2006, ApJ, 640, 1051

Grieves, N., Ge, J., Thomas, N., Ma, B., Sithajan, S., Ghezzi, L., Kimock, B., Willis, K.,

De Lee, N., Lee, B., Fleming, S. W., Agol, E., Troup, N., Paegert, M., Schneider, D. P.,

Stassun, K., Varosi, F., Zhao, B., Jian, L., Li, R., Porto de Mello, G. F., Bizyaev, D.,

Pan, K., Dutra-Ferreira, L., Lorenzo-Oliveira, D., Santiago, B. X., da Costa, L. N., Maia,

M. A. G., Ogando, R. L. C., & del Peloso, E. F. 2017, MNRAS, 467, 4264

Guo, X., Johnson, J. A., Mann, A. W., Kraus, A. L., Curtis, J. L., & Latham, D. W. 2017,

ApJ, 838, 25

Haisch, Karl E., J., Lada, E. A., & Lada, C. J. 2001, ApJ, 553, L153

Han, E., Wang, S. X., Wright, J. T., Feng, Y. K., Zhao, M., Fakhouri, O., Brown, J. I., &

Hancock, C. 2014, PASP, 126, 827

Hasegawa, Y. & Pudritz, R. E. 2012, ApJ, 760, 117

Heller, R. 2019, A&A, 628, A42

Heller, R. & Pudritz, R. 2015, ApJ, 806, 181

Heller, R., Williams, D., Kipping, D., Limbach, M. A., Turner, E., Greenberg, R., Sasaki,

T., Bolmont, E.,´ Grasset, O., Lewis, K., Barnes, R., & Zuluaga, J. I. 2014, Astrobiology,

14, 798

Holman, M., Touma, J., & Tremaine, S. 1997, Nature, 386, 254

Howard, A. W. 2013, Science, 340, 572

39 Howard, A. W., Marcy, G. W., Bryson, S. T., Jenkins, J. M., Rowe, J. F., Batalha, N. M.,

Borucki, W. J., Koch, D. G., Dunham, E. W., Gautier, III, T. N., Van Cleve, J., Cochran,

W. D., Latham, D. W., Lissauer, J. J., Torres, G., Brown, T. M., Gilliland, R. L.,

Buchhave, L. A., Caldwell, D. A., Christensen-Dalsgaard, J., Ciardi, D., Fressin, F.,

Haas, M. R., Howell, S. B., Kjeldsen, H., Seager, S., Rogers, L., Sasselov, D. D., Steffen,

J. H., Basri, G. S., Charbonneau, D., Christiansen, J., Clarke, B., Dupree, A., Fabrycky,

D. C., Fischer, D. A., Ford, E. B., Fortney, J. J., Tarter, J., Girouard, F. R., Holman,

M. J., Johnson, J. A., Klaus, T. C., Machalek, P., Moorhead, A. V., Morehead, R. C.,

Ragozzine, D., Tenenbaum, P., Twicken, J. D., Quinn, S. N., Isaacson, H., Shporer, A.,

Lucas, P. W., Walkowicz, L. M., Welsh, W. F., Boss, A., Devore, E., Gould, A., Smith,

J. C., Morris, R. L., Prsa, A., Morton, T. D., Still, M., Thompson, S. E., Mullally, F.,

Endl, M., & MacQueen, P. J. 2012, ApJS, 201, 15

Howarth, I. D. 2016, MNRAS, 457, 3769

Howell, S. B., Sobeck, C., Haas, M., Still, M., Barclay, T., Mullally, F., Troeltzsch, J.,

Aigrain, S., Bryson, S. T., Caldwell, D., Chaplin, W. J., Cochran, W. D., Huber, D.,

Marcy, G. W., Miglio, A., Najita, J. R., Smith, M., Twicken, J. D., & Fortney, J. J.

2014, PASP, 126, 398

Huang, C., Wu, Y., & Triaud, A. H. M. J. 2016, ApJ, 825, 98

Huerta, M., Johns-Krull, C. M., Prato, L., Hartigan, P., & Jaffe, D. T. 2008, ApJ, 678, 472

Ireland, M. J. & Kraus, A. L. 2014, in IAU Symposium, Vol. 299, Exploring the Formation

and Evolution of Planetary Systems, ed. M. Booth, B. C. Matthews, & J. R. Graham,

199–203

Itoh, Y., Hayashi, M., Tamura, M., Tsuji, T., Oasa, Y., Fukagawa, M., Hayashi, S. S., Naoi,

T., Ishii, M., Mayama, S., Morino, J.-i., Yamashita, T., Pyo, T.-S., Nishikawa, T., Usuda,

T., Murakawa, K., Suto, H., Oya, S., Takato, N., Ando, H., Miyama, S. M., Kobayashi,

N., & Kaifu, N. 2005, ApJ, 620, 984

40 Jackson, B., Arras, P., Penev, K., Peacock, S., & Marchant, P. 2017, ApJ, 835, 145

Johns-Krull, C. M., McLane, J. N., Prato, L., Crockett, C. J., Jaffe, D. T., Hartigan, P. M.,

Beichman, C. A., Mahmud, N. I., Chen, W., Skiff, B. A., Cauley, P. W., Jones, J. A., &

Mace, G. N. 2016a, ApJ, 826, 206

Johns-Krull, C. M., Prato, L., McLane, J. N., Ciardi, D. R., van Eyken, J. C., Chen, W.,

Stauffer, J. R., Beichman, C. A., Frazier, S. A., Boden, A. F., Morales-Calder´on,M., &

Rebull, L. M. 2016b, ApJ, 830, 15

Johnson, J. A., Aller, K. M., Howard, A. W., & Crepp, J. R. 2010, PASP, 122, 905

Johnson, J. A., Winn, J. N., Albrecht, S., Howard, A. W., Marcy, G. W., & Gazak, J. Z.

2009, PASP, 121, 1104

Kaib, N. A., Raymond, S. N., & Duncan, M. 2013, Nature, 493, 381

Kalas, P., Graham, J. R., & Clampin, M. 2005, Nature, 435, 1067

Kirkpatrick, J. D., Reid, I. N., Liebert, J., Cutri, R. M., Nelson, B., Beichman, C. A., Dahn,

C. C., Monet, D. G., Gizis, J. E., & Skrutskie, M. F. 1999, ApJ, 519, 802

Kirkpatrick, J. D., Reid, I. N., Liebert, J., Gizis, J. E., Burgasser, A. J., Monet, D. G.,

Dahn, C. C., Nelson, B., & Williams, R. J. 2000, AJ, 120, 447

Knutson, H. A., Fulton, B. J., Montet, B. T., Kao, M., Ngo, H., Howard, A. W., Crepp,

J. R., Hinkley, S., Bakos, G. A.,´ Batygin, K., Johnson, J. A., Morton, T. D., & Muirhead,

P. S. 2014, ApJ, 785, 126

Konishi, M., Hashimoto, J., & Hori, Y. 2018, ApJ, 859, L28

Kraus, A. L. & Ireland, M. J. 2012, ApJ, 745, 5

Kruijer, T. S., Burkhardt, C., Budde, G., & Kleine, T. 2017, Proceedings of the National

Academy of Science, 114, 6712

41 Lambrechts, M. & Johansen, A. 2012, A&A, 544, A32

Lin, D. N. C., Bodenheimer, P., & Richardson, D. C. 1996, Nature, 380, 606

Lin, D. N. C. & Papaloizou, J. 1986, ApJ, 309, 846

Lineweaver, C. H. 2001, , 151, 307

Lissauer, J. J., Hubickyj, O., D’Angelo, G., & Bodenheimer, P. 2009, , 199, 338

Liu, M. C., Magnier, E. A., Deacon, N. R., Allers, K. N., Dupuy, T. J., Kotson, M. C.,

Aller, K. M., Burgett, W. S., Chambers, K. C., Draper, P. W., Hodapp, K. W., Jedicke,

R., Kaiser, N., Kudritzki, R. P., Metcalfe, N., Morgan, J. S., Price, P. A., Tonry, J. L.,

& Wainscoat, R. J. 2013, ApJ, 777, L20

Lufkin, G., Quinn, T., Wadsley, J., Stadel, J., & Governato, F. 2004, MNRAS, 347, 421

Luhman, K. L., Wilson, J. C., Brandner, W., Skrutskie, M. F., Nelson, M. J., Smith, J. D.,

Peterson, D. E., Cushing, M. C., & Young, E. 2006, ApJ, 649, 894

Ma, B. & Ge, J. 2014, MNRAS, 439, 2781

Mahmud, N. I., Crockett, C. J., Johns-Krull, C. M., Prato, L., Hartigan, P. M., Jaffe, D. T.,

& Beichman, C. A. 2011, ApJ, 736, 123

Mann, A. W., Gaidos, E., Mace, G. N., Johnson, M. C., Bowler, B. P., LaCourse, D.,

Jacobs, T. L., Vanderburg, A., Kraus, A. L., Kaplan, K. F., & Jaffe, D. T. 2016a, ApJ,

818, 46

Mann, A. W., Newton, E. R., Rizzuto, A. C., Irwin, J., Feiden, G. A., Gaidos, E., Mace,

G. N., Kraus, A. L., James, D. J., Ansdell, M., Charbonneau, D., Covey, K. R., Ireland,

M. J., Jaffe, D. T., Johnson, M. C., Kidder, B., & Vanderburg, A. 2016b, AJ, 152, 61

Marcy, G., Butler, R. P., Fischer, D., Vogt, S., Wright, J. T., Tinney, C. G., & Jones,

H. R. A. 2005, Progress of Theoretical Physics Supplement, 158, 24

42 Marcy, G. W. & Butler, R. P. 2000, PASP, 112, 137

Marcy, G. W., Butler, R. P., Vogt, S. S., Liu, M. C., Laughlin, G., Apps, K., Graham,

J. R., Lloyd, J., Luhman, K. L., & Jayawardhana, R. 2001, ApJ, 555, 418

Marois, C., Macintosh, B., Barman, T., Zuckerman, B., Song, I., Patience, J., Lafreni`ere,

D., & Doyon, R. 2008, Science, 322, 1348

Marois, C., Zuckerman, B., Konopacky, Q. M., Macintosh, B., & Barman, T. 2010, Nature,

468, 1080

Mart´ın,E. L., Brandner, W., Bouvier, J., Luhman, K. L., Stauffer, J., Basri, G., Zapatero

Osorio, M. R., & Barrado y Navascu´es, D. 2000, ApJ, 543, 299

Masuda, K. 2017, AJ, 154, 64

Matsuyama, I., Johnstone, D., & Murray, N. 2003, ApJ, 585, L143

Mayer, L., Quinn, T., Wadsley, J., & Stadel, J. 2004, ApJ, 609, 1045

Mayor, M., Marmier, M., Lovis, C., Udry, S., S´egransan,D., Pepe, F., Benz, W., Bertaux,

J. ., Bouchy, F., Dumusque, X., Lo Curto, G., Mordasini, C., Queloz, D., & Santos, N. C.

2011, ArXiv e-prints

Mayor, M. & Queloz, D. 1995, Nature, 378, 355

McKee, C. F. & Ostriker, E. C. 2007, ARA&A, 45, 565

Mizuno, H., Nakazawa, K., & Hayashi, C. 1980, Earth and Planetary Science Letters, 50,

202

Mordasini, C., Alibert, Y., Klahr, H., & Henning, T. 2012, A&A, 547, A111

Mordasini, C., Molli`ere,P., Dittkrist, K. M., Jin, S., & Alibert, Y. 2015, International

Journal of Astrobiology, 14, 201

Murray-Clay, R. A., Chiang, E. I., & Murray, N. 2009, ApJ, 693, 23

43 Nagasawa, M., Ida, S., & Bessho, T. 2008, ApJ, 678, 498

Naoz, S., Farr, W. M., Lithwick, Y., Rasio, F. A., & Teyssandier, J. 2011, Nature, 473, 187

Naoz, S., Farr, W. M., & Rasio, F. A. 2012, ApJ, 754, L36

Nayakshin, S. 2011, MNRAS, 416, 2974

—. 2017, , 34, e002

Nelson, R. P., Papaloizou, J. C. B., Masset, F., & Kley, W. 2000, MNRAS, 318, 18

Ngo, H., Knutson, H. A., Hinkley, S., Bryan, M., Crepp, J. R., Batygin, K., Crossfield, I.,

Hansen, B., Howard, A. W., Johnson, J. A., Mawet, D., Morton, T. D., Muirhead, P. S.,

& Wang, J. 2016, ApJ, 827, 8

Onitsuka, M., Fukui, A., Narita, N., Hirano, T., Kusakabe, N., Ryu, T., & Tamura, M.

2017, ArXiv e-prints

Owen, J. E. & Wu, Y. 2013, ApJ, 775, 105

Perryman, M. A. C., Brown, A. G. A., Lebreton, Y., Gomez, A., Turon, C., Cayrel de

Strobel, G., Mermilliod, J. C., Robichon, N., Kovalevsky, J., & Crifo, F. 1998, A&A,

331, 81

Persson, C. M., Csizmadia, S., Mustill, A. e. J., Fridlund, M., Hatzes, A. P., Nowak,

G., Georgieva, I., Gandolfi, D., Davies, M. B., Livingston, J. H., Palle, E., Monta˜nes

Rodr´ıguez,P., Endl, M., Hirano, T., Prieto-Arranz, J., Korth, J., Grziwa, S., Esposito,

M., Albrecht, S., Johnson, M. C., Barrag´an, O., Parviainen, H., Van Eylen, V., Alonso

Sobrino, R., Beck, P. G., Cabrera, J., Carleo, I., Cochran, W. D., Dai, F., Deeg, H. J., de

Leon, J. P., Eigm¨uller,P., Erikson, A., Fukui, A., Gonz´alez-Cuesta,L., Guenther, E. W.,

Hidalgo, D., Hjorth, M., Kabath, P., Knudstrup, E., Kusakabe, N., Lam, K. W. F., Lund,

M. N., Luque, R., Mathur, S., Murgas, F., Narita, N., Nespral, D., Niraula, P., Olofsson,

A. O. H., P¨atzold,M., Rauer, H., Redfield, S., Ribas, I., Skarka, M., Smith, A. M. S.,

Subjak, J., & Tamura, M. 2019, A&A, 628, A64

44 Petrovich, C. & Tremaine, S. 2016, ApJ, 829, 132

Pollack, J. B., Hubickyj, O., Bodenheimer, P., Lissauer, J. J., Podolak, M., & Greenzweig,

Y. 1996, , 124, 62

Quinn, S. N., White, R. J., Latham, D. W., Buchhave, L. A., Torres, G., Stefanik, R. P.,

Berlind, P., Bieryla, A., Calkins, M. C., Esquerdo, G. A., F˝ur´esz,G., Geary, J. C., &

Szentgyorgyi, A. H. 2014, ApJ, 787, 27

Rafikov, R. R. 2005, ApJ, 621, L69

—. 2006, ApJ, 648, 666

Rasio, F. A. & Ford, E. B. 1996, Science, 274, 954

Raymond, S. N., Barnes, R., & Kaib, N. A. 2006, ApJ, 644, 1223

Raymond, S. N. & Izidoro, A. 2017, , 297, 134

Raymond, S. N., Quinn, T., & Lunine, J. I. 2005, , 177, 256

Reipurth, B. & Clarke, C. 2001, AJ, 122, 432

Rice, W. K. M. & Armitage, P. J. 2003, ApJ, 598, L55

Sallum, S., Follette, K. B., Eisner, J. A., Close, L. M., Hinz, P., Kratter, K., Males, J.,

Skemer, A., Macintosh, B., Tuthill, P., Bailey, V., Defr`ere, D., Morzinski, K., Rodigas,

T., Spalding, E., Vaz, A., & Weinberger, A. J. 2015, Nature, 527, 342

Santerne, A., Moutou, C., Tsantaki, M., Bouchy, F., H´ebrard, G., Adibekyan, V.,

Almenara, J. M., Amard, L., Barros, S. C. C., Boisse, I., Bonomo, A. S., Bruno, G.,

Courcol, B., Deleuil, M., Demangeon, O., D´ıaz,R. F., Guillot, T., Havel, M., Montagnier,

G., Rajpurohit, A. S., Rey, J., & Santos, N. C. 2016, A&A, 587, A64

Santos, N. C., Israelian, G., & Mayor, M. 2004, A&A, 415, 1153

45 Sato, B., Fischer, D. A., Henry, G. W., Laughlin, G., Butler, R. P., Marcy, G. W., Vogt,

S. S., Bodenheimer, P., Ida, S., Toyota, E., Wolf, A., Valenti, J. A., Boyd, L. J., Johnson,

J. A., Wright, J. T., Ammons, M., Robinson, S., Strader, J., McCarthy, C., Tah, K. L.,

& Minniti, D. 2005, ApJ, 633, 465

Schlaufman, K. C. 2018, ApJ, 853, 37

Shahaf, S. & Mazeh, T. 2019, MNRAS, 487, 3356

Sinukoff, E., Howard, A. W., Petigura, E. A., Fulton, B. J., Isaacson, H., Weiss, L. M.,

Brewer, J. M., Hansen, B. M. S., Hirsch, L., Christiansen, J. L., Crepp, J. R., Crossfield,

I. J. M., Schlieder, J. E., Ciardi, D. R., Beichman, C. A., Knutson, H. A., Benneke, B.,

Dressing, C. D., Livingston, J. H., Deck, K. M., L´epine, S., & Rogers, L. A. 2017, AJ,

153, 70

Strom, S. E., Edwards, S., & Skrutskie, M. F. 1993, in Protostars and Planets III, ed. E. H.

Levy & J. I. Lunine, 837

Takeuchi, T., Miyama, S. M., & Lin, D. N. C. 1996, ApJ, 460, 832

Tanaka, H., Takeuchi, T., & Ward, W. R. 2002, ApJ, 565, 1257

Triaud, A. H. M. J., Collier Cameron, A., Queloz, D., Anderson, D. R., Gillon, M., Hebb, L.,

Hellier, C., Loeillet, B., Maxted, P. F. L., Mayor, M., Pepe, F., Pollacco, D., S´egransan,

D., Smalley, B., Udry, S., West, R. G., & Wheatley, P. J. 2010, A&A, 524, A25

Trilling, D. E., Benz, W., Guillot, T., Lunine, J. I., Hubbard, W. B., & Burrows, A. 1998,

ApJ, 500, 428

Udry, S., Mayor, M., Naef, D., Pepe, F., Queloz, D., Santos, N. C., & Burnet, M. 2002,

A&A, 390, 267

Udry, S. & Santos, N. C. 2007, ARA&A, 45, 397

46 Valenti, J. & Fischer, D. 2008, Astronomical Society of the Pacific Conference Series, Vol.

384, Stellar Metallicity and Planet Formation, ed. G. van Belle, 292 van Eyken, J. C., Ciardi, D. R., von Braun, K., Kane, S. R., Plavchan, P., Bender, C. F.,

Brown, T. M., Crepp, J. R., Fulton, B. J., Howard, A. W., Howell, S. B., Mahadevan, S.,

Marcy, G. W., Shporer, A., Szkody, P., Akeson, R. L., Beichman, C. A., Boden, A. F.,

Gelino, D. M., Hoard, D. W., Ram´ırez,S. V., Rebull, L. M., Stauffer, J. R., Bloom, J. S.,

Cenko, S. B., Kasliwal, M. M., Kulkarni, S. R., Law, N. M., Nugent, P. E., Ofek, E. O.,

Poznanski, D., Quimby, R. M., Walters, R., Grillmair, C. J., Laher, R., Levitan, D. B.,

Sesar, B., & Surace, J. A. 2012, ApJ, 755, 42

Wang, J. & Fischer, D. A. 2015, AJ, 149, 14

Ward, W. R. 1997, ApJ, 482, L211

Winn, J. N., Fabrycky, D., Albrecht, S., & Johnson, J. A. 2010, ApJ, 718, L145

Wright, J. T., Marcy, G. W., Howard, A. W., Johnson, J. A., Morton, T. D., & Fischer,

D. A. 2012, ApJ, 753, 160

Wu, D.-H., Wang, S., Zhou, J.-L., Steffen, J. H., & Laughlin, G. 2018, AJ, 156, 96

Wu, Y. & Murray, N. 2003, ApJ, 589, 605

Yu, L., Donati, J.-F., H´ebrard,E. M., Moutou, C., Malo, L., Grankin, K., Hussain, G.,

Collier Cameron, A., Vidotto, A. A., Baruteau, C., Alencar, S. H. P., Bouvier, J., Petit,

P., Takami, M., Herczeg, G., Gregory, S. G., Jardine, M., Morin, J., M´enard,F., &

Matysse Collaboration. 2017, MNRAS

Yu, L., Winn, J. N., Gillon, M., Albrecht, S., Rappaport, S., Bieryla, A., Dai, F., Delrez,

L., Hillenbrand, L., Holman, M. J., Howard, A. W., Huang, C. X., Isaacson, H., Jehin,

E., Lendl, M., Montet, B. T., Muirhead, P., Sanchis-Ojeda, R., & Triaud, A. H. M. J.

2015, ApJ, 812, 48

47 Zhou, G., Huang, C. X., Bakos, G. A.,´ Hartman, J. D., Latham, D. W., Quinn, S. N.,

Collins, K. A., Winn, J. N., Wong, I., Kov´acs,G., Csubry, Z., Bhatti, W., Penev, K.,

Bieryla, A., Esquerdo, G. A., Berlind, P., Calkins, M. L., de Val-Borro, M., Noyes,

R. W., L´az´ar,J., Papp, I., S´ari,P., Kov´acs,T., Buchhave, L. A., Szklenar, T., B´eky,

B., Johnson, M. C., Cochran, W. D., Kniazev, A. Y., Stassun, K. G., Fulton, B. J.,

Shporer, A., Espinoza, N., Bayliss, D., Everett, M., Howell, S. B., Hellier, C., Anderson,

D. R., Collier Cameron, A., West, R. G., Brown, D. J. A., Schanche, N., Barkaoui, K.,

Pozuelos, F., Gillon, M., Jehin, E., Benkhaldoun, Z., Daassou, A., Ricker, G., Vand

erspek, R., Seager, S., Jenkins, J. M., Lissauer, J. J., Armstrong, J. D., Collins, K. I.,

Gan, T., Hart, R., Horne, K., Kielkopf, J. F., Nielsen, L. D., Nishiumi, T., Narita, N.,

Palle, E., Relles, H. M., Sefako, R., Tan, T. G., Davies, M., Goeke, R. F., Guerrero, N.,

Haworth, K., & Villanueva, S. 2019, AJ, 158, 141

Zuckerman, B. 2014, ApJ, 791, L27

48 Chapter 2

Survey Overview

This chapter gives an overview of the spectroscopic survey for finding young substellar companions around pre-main sequence stars and characterizing stellar parameters

(projected rotational velocity, effective temperature, rotation period, and limit on the radius). The infrared survey is built on an initial optical survey of pre-main sequence stars that began in 2004 (Huerta et al. 2008; Prato et al. 2008; Mahmud et al. 2011;

Crockett et al. 2011, 2012). The current RV survey primarily relies on infrared spectroscopy, however, optical spectroscopy is used to supplement the stellar characterization results, and ground-based visual-light photometry and K2 photometry are used to determine stellar rotation periods, which are needed to rule out stellar activity as the source of RV variability indicating a substellar companion detection. I conducted and analyzed the infrared spectroscopic survey using IGRINS data. Collaborators Chris Johns-Krull and Adolfo

Carvalho analyzed and provided the optical spectroscopy results. Brian Skiff and Lauren

Biddle analyzed and provided the ground-based photometry and space-based photometry, respectively. This section discusses the sample selection, survey design, observations and data reductions.

2.1 Sample Selection

The targets for the RV survey were primarily selected from Herbig & Bell (1988), a catalog that lists 742 pre-main sequence stars observed and characterized with various

49 spectrographs. The sample was chosen based on 3 main criteria: 1) brightness (V<15),

2) slow rotation (v sin i <20 km s−1), 3) spectral type (K and M stars). Brighter targets

allow for more efficient observations, and most of the targets have a K band magnitude

∼7-9 (Figure 2.1). Slower stellar rotational velocities ensure that the spectra have deep,

narrow absorption lines, which are important for accurate RV analysis. Furthermore, K

and M stars have maximum signal-to-noise ratio (S/N) in the red and infrared part of the

spectrum. Pre-main sequence stars of late K and early M spectral types will evolve to

become late G or early K solar-type stars. A correlation between giant planet occurrence

and stellar mass has been identified (Johnson et al. 2010), whereas systems with massive

disks are able to form both high-mass stars and planets. Therefore, these stars offer a

compromise between having deep CO lines for accurate RV analysis and being massive

enough to offer a reasonable chance of hosting a giant planet.

Additional stars, which did not necessarily meet the first set of criteria and are either

fainter, faster rotators, or have higher mass, were also included in the sample. Ideally, a

RV survey would include only single stars to avoid systems with stellar companion induced

RV motion or spectral contamination. However, the majority of young stars reside in

binary systems (Ghez et al. 1993; Simon et al. 1995) and Kraus et al. (2011) found that

∼65-75% of all Taurus members are in multiple systems. Therefore, in order to increase the

sample size, several binaries with angular separations >0.0500, corresponding to wide enough

physical separations that they don’t induce RV variation over the timescale of observations,

were also included in the sample. Furthermore, close-in giant planets and brown dwarfs

may preferentially be found in binary systems (Zucker & Mazeh 2002; Udry & Santos 2007;

Ngo et al. 2016; Fontanive et al. 2019). The selected binaries are wide enough that RV

variability from a stellar companion is seen over tens of years or longer, and therefore do

not impact the search for short-period variability. About 45% of the final survey sample are

in binary or multiple systems, and when resolvable, the primary star was targeted because

there is evidence that protoplanetary disks in binary systems are more likely to form and

persist around the primary than the secondary (Monin et al. 2007). The stellar companion

50 Figure 2.1 A histogram illustrating the K band magnitudes of the 70 pre-main sequence stars in the sample. Most of the targets are relatively bright with a magnitude between 7 and 9.

51 was not resolved for about 2/3 of the multiple systems in the sample, leading to potential light contamination from the secondary. Several additional targets were also drawn from the sample identified by the NASA Space Interferometry Mission (started in 1998 and cancelled in 2010), which aimed to search for young planetary systems (Beichman et al. 2002).

The above selection process resulted in 142 young stars of spectral type F-M. The targets are primarily located in the Taurus-Auriga star-forming region with age <5 Myr old (Kraus

& Hillenbrand 2009), 14 targets are part of the Pleiades open cluster (∼100 Myr old; Meynet et al. 1993), and 3 are in the Hyades cluster (∼600 Myr old; Perryman et al. 1998). The survey sample was adjusted to include only the youngest targets in the relatively nearby

+12 Taurus star-forming region (145−16 pc; Yan et al. 2019); the vast majority of low-mass stars in Taurus have ages <3 Myr (Brice˜noet al. 2002), although an older subpopulation also exists in this region (&10 Myr; Daemgen et al. 2015; Kraus et al. 2017; Luhman 2018; Zhang et al. 2018). The sample was further reduced to only include the low-mass pre-main sequence stars of estimated spectral type K and M. The final number of targets considered for stellar characterization was 70.

About 70% of the pre-main sequence star sample show evidence of having an accreting circumstellar disk based on the strength of the Hα emission lines, and are thus classified as

classical T Tauri stars (CTTSs), while the remaining targets that show no evidence of such

a disk are classified as weak-line T Tauri stars (WTTSs) (Herbig & Bell 1988; Andrews &

Williams 2005; McCabe et al. 2006; Monin et al. 2010; Kraus et al. 2012; Schaefer et al.

2012; P´ericaudet al. 2017; Akeson et al. 2019). This percentage is expected for a population

of young stars; about 60-80% of stars younger than 2 Myr still host protoplanetary disks

(Fedele et al. 2010). Protoplanets form in the circumstellar disk environments of CTTSs

within <0.1-1 Myr (Najita & Kenyon 2014; Manara et al. 2018), so these stars could offer

a view of the early stages of planet formation.

Ten targets were chosen from the sample of 70 pre-main sequence stars for RV follow-up

to search for substellar companions. These targets were chosen either because of published

candidate detections presented in the literature (CI Tau, Johns-Krull et al. 2016; V830 Tau,

52 Donati et al. 2016) or RV variability in optical spectroscopic data (DK Tau) or infrared spectroscopic data (V1075 Tau). The remaining follow-up targets were chosen based on photometric data from the K2 space mission (AA Tau, DM Tau, GI Tau, GM Tau, IQ Tau,

LkCa 15). These targets are CTTSs and were identified as objects of interest for a possible study of star-planet-disk interactions with a collaborator, Lauren Biddle (see Biddle et al.

2018 for details).

Table 2.1 lists the targets in the sample, along with their positions, K band magnitudes, and number of infrared and optical spectroscopic observations. A color-magnitude diagram and Hertzsprung-Russell diagram of these pre-main sequence stars are shown in Figure 2.2 and Figure 2.3, respectively. H and K magnitudes and parallaxes, from which distances and absolute magnitudes could be calculated, were collected from the Simbad database1 for

∼70% of the sample. Effective temperatures were measured using line-depth ratios by a collaborator, Ricardo Lopez-Valdivia (see Chapter 3 for details).

Table 2.1: Sample of 70 Pre-Main Sequence Stars

Target RA Dec (J2000) K N obs (IR) N obs (opt)b

AA Taua 04 34 55 +24 28 53 8.05 31 91

CI Taua 04 33 52 +22 50 30 7.79 75 231

DK Taua 04 30 44 +26 01 24 7.10 54 191

DM Taua 04 33 48 +18 10 09 9.52 16 ···

GI Taua 04 33 34 +24 21 17 7.89 22 21

GM Taua 04 38 21 +26 09 14 10.63 18 ···

IQ Taua 04 29 51 +26 06 44 7.78 25 161

LkCa 15a 04 39 17 +22 21 03 8.16 30 101

V830 Taua 04 33 10 +24 33 43 8.42 88 121

V1075 Taua 04 32 09 +17 57 23 8.85 93 211

Continued on next page

1http://simbad.u-strasbg.fr

53 Target RA Dec (J2000) K N obs (IR) N obs (opt)b

BP Tau 04 19 15 +29 06 26 7.74 7 421

CoKu LkHa 332 G2 04 42 05 +25 22 56 8.23 4 ···

CoKu Tau 3 04 35 40 +24 11 08 8.41 4 ···

CW Tau 04 14 17 +28 10 57 7.13 6 21

CZ Tau 04 18 31 +28 16 58 9.36 5 ···

DE Tau 04 21 55 +27 55 06 7.80 6 ···

DG Tau 04 27 04 +26 06 16 6.99 3 101

DH Tau 04 29 40 +26 32 58 8.18 22 81

DI Tau 04 29 42 +26 32 49 8.39 5 101

DL Tau 04 33 39 +25 20 38 7.96 7 71

DN Tau 04 35 27 +24 14 58 8.02 3 431

DR Tau 04 47 06 +16 58 42 6.87 5 101

DS Tau 04 47 48 +29 25 11 8.04 19 161

FM Tau 04 14 13 +28 12 49 8.76 6 ···

FZ Tau 04 32 31 +24 20 03 7.35 5 ···

GG Tau A 04 32 30 +17 31 41 7.36 4 ···

GK Tau 04 33 34 +24 21 05 7.47 5 151

GM Aur 04 55 10 +30 21 59 8.23 9 121

GN Tau 04 39 20 +25 45 02 8.06 5 ···

GV Tau 04 29 23 +24 33 00 8.05 4 ···

Haro 6-37 04 46 58 +17 02 38 ··· 4 11

HK Tau 04 31 50 +24 24 18 8.59 5 ···

HN Tau 04 33 39 +17 51 52 8.38 4 ···

HO Tau 04 35 20 +22 32 14 10.24 4 ···

IP Tau 04 24 57 +27 11 56 8.35 6 11

IRAS F04113+2758 04 14 26 +28 06 05 ··· 4 ···

Continued on next page

54 Target RA Dec (J2000) K N obs (IR) N obs (opt)b

IRAS F04192+2647 04 22 18 +26 54 00 9.01 3 ···

IRAS F04248+2612 04 27 57 +26 19 18 11.03 4 ···

IRAS F04325+2402 04 35 35 +24 08 19 11.60 4 ···

IRAS F04370+2559 04 40 08 +26 05 25 8.87 4 ···

IRAS F04385+2550 04 41 39 +25 56 27 9.20 3 ···

IT Tau 04 33 54 +26 13 27 7.86 4 ···

IW Tau 04 41 04 +24 51 06 8.28 7 321

L1551 IRS 5 04 31 34 +18 08 05 9.26 4 ···

L1551-55 04 32 43 +18 02 56 9.31 5 ···

L1642-2 04 34 49 -14 13 08 ··· 3 ···

LkCa 4 04 16 28 +28 07 35 8.32 7 91

LkCa 5 04 17 38 +28 33 00 9.05 6 21

LkHa 332/G1 04 42 07 +25 23 03 7.95 5 ···

NTTS 040142+2150 NE 04 04 39 +21 58 21 10.10 3 ···

NTTS 040142+2150 SW 04 04 39 +21 58 18 9.97 3 ···

NTTS 040234+2143 04 05 30 +21 51 10 10.06 3 ···

NTTS 041559+1716 04 18 51 +17 23 16 9.27 18 31

NTTS 043230+1746 04 35 24 +17 51 42 9.08 3 ···

RW Aur 05 07 49 +30 24 05 7.02 4 32

RX J0425.3+2618 04 25 18 +26 17 50 9.00 3 ···

T Tau 04 21 59 +19 32 06 5.33 5 ···

UX Tau B 04 30 04 +18 13 50 ··· 5 ···

V410 Tau 04 18 31 +28 27 16 7.63 4 31

V710 Tau A 04 31 58 +18 21 37 ··· 3 ···

V710 Tau B 04 31 58 +18 21 37 ··· 3 ···

V807 Tau 04 33 06 +24 09 55 6.96 4 ···

Continued on next page

55 Target RA Dec (J2000) K N obs (IR) N obs (opt)b

V819 Tau 04 19 26 +28 26 14 8.42 5 ···

V836 Tau 05 03 06 +25 23 19 8.56 6 131

V928 Tau 04 32 18 +24 22 27 8.11 4 ···

V1095 Tau 04 13 14 +28 19 10 8.63 4 ···

V1096 Tau 04 13 27 +28 16 24 7.46 4 51

V1115 Tau 04 36 19 +25 42 59 8.58 3 151

XZ Tau 04 31 40 +18 13 57 7.29 6 51

ZZ Tau 04 30 51 +24 42 22 8.44 4 ···

(a) RV follow-up target (b) McDonald Observatory 1) 107-inch 2) 82-inch

2.2 Infrared Spectroscopy Survey

2.2.1 Finding Young Substellar Companions: Challenges and Solutions

An infrared survey can address multiple challenges involved in detecting young planets.

Pre-main sequence stars are affected by dust extinction and are found at large distances

(>100 pc); they are therefore inherently faint. The targets chosen for this survey are in the relatively nearby Taurus-Auriga region, with typical K band magnitudes between 7-9.

The survey is comprised of low-mass stars of spectral type K and M, which are brightest at near-infrared wavelengths (Bean et al. 2010; Blake et al. 2010).

Due to their strong kilogauss magnetic fields (Johns-Krull 2007), T Tauri stars are highly magnetically active. Starspots in particular are a significant source of RV variation that can mimic a planet detection signal, especially when the data is unevenly sampled

(Saar & Donahue 1997; Desort et al. 2007; Reiners et al. 2010; Rajpaul et al. 2016). As starspots on the star’s surface rotate in and out of view of the observer, they block some of the stellar flux and create an imbalance between the redshifted and blueshifted sides of the star, thus causing distortions and asymmetries in the spectral absorption lines (Saar

56 Figure 2.2 A color-magnitude diagram with the 50 pre-main sequence stars in the final sample that have measurements of H and K magnitudes and parallaxes. This sample includes the 70 general survey targets and the 10 RV follow-up targets. The average K band magnitude of these targets is 8.4, and the average distance is ∼138 pc. The Pleiades, which are on the zero age main sequence (the phase in a star’s evolution when it first joins the main sequence by fusing hydrogen in its core), are overplotted in gray to illustrate the evolutionary stage of the pre-main sequence star sample. The Pleiades H and K magnitudes were measured by Bouy et al. (2015) and the Gaia DR2 distance estimates were determined by Abramson (2018). Typical error bars are shown in the upper left (the absolute K- magnitude error is typically ∼0.01, and therefore too small to appear on the figure; the typical error on H-K is ∼0.03).

57 Figure 2.3 A Hertzsprung-Russell diagram of the 46 pre-main sequence stars in the final sample that have measurements of H and K magnitudes, parallaxes, and effective temperatures (see Section 3.2.1 for a description of the temperature calculations). This sample includes the general survey targets and follow-up RV targets. The average effective temperature for this sample is 3800 K. The Pleiades are overplotted in gray to demonstrate the evolutionary stage of the pre-main sequence star sample relative to the zero age main sequence. Typical error bars are shown in the upper left (the absolute K-magnitude error is typically ∼0.01, and therefore too small to appear on the figure; the error on the effective temperatures is 200 K).

58 Figure 2.4 A diagram from Haywood (2016) illustrating how starspots cause RV variability that can mimic a planet detection signal. As the starspot rotates on the stellar surface, it blocks flux from the redshifted and blueshifted halves of the star. This creates asymmetries in the spectral line profile that shifts the centroid and translates to a variation in the measured RVs. While this example shows the effect of a small starspot, comparable to those found on the Sun, pre-main sequence stars typically have heightened activity leading to larger, more frequent, and stable starspots.

& Donahue 1997; Haywood 2016). The diagram shown in Figure 2.4 demonstrates this effect. Starspots on young stars are also stable, long-lived, and capable of persisting for hundreds of days or even several years (Hall & Henry 1994; Hu´elamoet al. 2008; Mahmud et al. 2011). They may also cover a large percentage of the star’s surface, possibly up to

80% (Gully-Santiago et al. 2017). In a couple cases, young star RV variability has been mistaken for a Jupiter-like planet (Setiawan et al. 2008; Hern´an-Obispo et al. 2010), but subsequent observations indicated that these were misidentifications caused by starspots

(Hu´elamoet al. 2008; Figueira et al. 2010).

Faculae are hotspots on the stellar surface caused by magnetic activity (Spruit 1976).

Like starspots, they also cause significant RV fluctuation (Haywood 2016). However studies on the variability of Sun-like stars by Radick et al. (1998) and Lockwood et al. (2007) suggest that photometric variations of young stars tend to be anti-correlated with chromospheric

59 variations, suggesting that their surfaces are dominated by starspots during periods of high stellar activity levels, and not faculae. Hu´elamo et al. (2008) predict that the RV variations caused by hotspots on young stars would be significantly weaker than those caused by starspots, based on their relatively small area coverage. Flares and coronal mass ejections, which lead to sudden increases in the star’s brightness, can also induce RV variation, but on a much shorter timescale than other stellar activity phenomena (Haywood 2016). Transient accretion hotspots may arise on the surface of a CTTS through star-disk interactions, and can also be detected as RV variability (Herbst et al. 1994; Johns-Krull et al. 2016).

Spot-induced RV variations exhibit a wavelength dependence, while companion-induced variability does not. Because of the lower contrast at longer wavelengths between a star’s surface temperature and starspot temperature, RV amplitudes caused by starspots will be decreased at infrared wavelengths, typically by a factor ∼2-5 (Mahmud et al. 2011; Crockett et al. 2012; Bailey et al. 2012; Carleo et al. 2018). This factor is expected to vary from star to star, however, since the magnitude of the decrease is dependent on the temperature contrast between the photosphere and starspots; the reduction in RV variability observed in the near-infrared is more pronounced when the temperature contrast is small (Reiners et al. 2010). This reduction effect is demonstrated in Figure 2.5. The dissimilar optical and infrared amplitudes indicate that large starspots are causing the RV modulation, rather than a planet detection. Matched RV amplitudes in the optical and infrared are indicative of planet detections (Johns-Krull et al. 2016).

Unfortunately, this wavelength dependence of cool spot-induced RV variations does not apply to RV variability caused by faculae since hotspots have a significantly higher temperature than the stellar photosphere (Hartigan et al. 1989; Basri & Batalha 1990;

Hartigan et al. 1991; Valenti et al. 1993). Accretion hotspots also induce both spectroscopic and photometric variability. Photometric observations may provide a simple test of this scenario (Haywood 2016; Johns-Krull et al. 2016). Photometry can indicate a false-positive by determining a stellar rotation period to compare to the modulation of the RV variation,

60 Figure 2.5 Phase folded RVs measured in optical and infrared wavelengths for Hubble I 4 from Mahmud et al. 2011. The optical (blue squares) and infrared (red diamonds) phase- folded RV amplitudes do not match, indicating the RV variability observed in this pre-main sequence star is likely the result of large star spots.

61 thereby inferring a source of the variability (i.e. stellar rotation and magnetic activity or gravitational effects from a companion).

The survey conducted in this dissertation is primarily based on the infrared spectroscopic data. Even in the K band, young stars can exhibit RV amplitudes on the order of hundreds of meters per second with stable phase coherence over several years. The survey also includes photometrically-derived stellar rotation periods. Stellar surface features, like cool and hot spots induce both spectroscopic and photometric variability that are modulated by the rotation period of the star, therefore the stellar rotation period can be used to remove these signals from the RV dataset. Infrared data alone are not sufficient to determine if RV signals are planetary in nature, but when combined with photometric rotation periods, it is possible to distinguish between starspots and the presence of a companion by comparing the stellar rotation period with the possible companion orbital period.

2.2.2 Observations and Data Reduction

This survey used the high resolution infrared spectrograph, IGRINS (Immersion GRating

INfrared Spectrograph), which was deployed as a visiting instrument to the 4.3-m Lowell

Discovery Telescope (LDT) for 6 months of the year over 3 years, starting in September

2016. Additional data were also used from previous years when IGRINS was at the 2.7-m

Harlan J. Smith Telescope at McDonald Observatory.

IGRINS can simultaneously observe the near-infrared H and K bands (43 total orders over 1.5-2.5 µm), at a spectral resolution of ∼45,000. It has a high sensitivity resulting from a straightforward design with no moving parts and simplified optics. IGRINS has a silicon immersion echelle grating as its primary disperser; the high refractive index of immersed silicon at near-infrared wavelengths (n∼3.4) allows IGRINS to have a large spectral grasp

(instantaneous spectral coverage) and yet a compact design. Individual volume phase holographic gratings serve as the cross-dispersing elements and separate the overlapping orders. IGRINS is cooled so the H and K detectors remain at ∼65 K so thermal noise is minimized, and it maintains a low readnoise (<5 e-). IGRINS is mounted on the Cassegrain

62 Table 2.2 IGRINS Specifications Wavelength Coverage H: 1.47-1.81µm, K: 1.95-2.48µm Spectral Resolution R = λ/∆λ ∼ 45, 000 Resolution Element (3.3 pixels) 3.84 × 10−5 µm ; ∼6.6 km/s Slit Scale on LDT 0.6300 × 9.300 Slit Viewer Camera Field-of-View ∼ 117.400 × 70.300 Detectors 2k × 2k Teledyne HgCdTe HAWAII-2RG CMOSs Detector Gain H: ∼ 2.05 e−/ADU, K: ∼ 2.21 e−/ADU Read Noise H: ∼ 10.92 e−, K: ∼ 8.93 e− focus of the telescope. Table 2.2 shows the specifications of the IGRINS instrument and

Figure 2.6 shows an example of an IGRINS echellogram. Additional discussion of the

IGRINS design and capabilities can be found in Yuk et al. (2010), Park et al. (2014), and

Mace et al. (2018).

During the follow-up observing season for the RV survey (fall/winter 2018-2019), the resolution in certain regions of the K band was reduced to ∼20,000, due to loose fasteners on the K band detector mount. This impacted the region of analysis used for the pre-main sequence stellar characterization and RV analysis (see the 2018 observation in Figure 3.1 compared to previous epochs). A total of 211 spectra of the ten follow-up targets were affected, which is about half (∼47%) of the RV observations.

A survey to find young substellar companions requires significant telescope resources.

Fischer et al. (2005) showed that hot Jupiters could be identified with only 3-4 observations taken over a few days because of their high RV amplitudes and short orbital periods, however, due to the high levels of stellar activity of pre-main sequence stars, intense monitoring and substantial data are needed (Mayor et al. 2011; Benatti 2018). Nightly observations over the course of a week, which corresponds to the typical rotation period of

T Tauri stars and the orbital period of hot Jupiters, is ideal, however, this was not often possible to obtain given limited telescope time allocations. Given the intensive demands of such a survey, the RV targets were limited to a subset of ten stars.

In total, the survey includes 963 spectra of the targets in the full sample and 452 total observations of the 10 RV targets. Table 2.3 summarizes the number of nights observed

63 Figure 2.6 IGRINS echellogram of simulated flats in H (left) and K (right) bands. Wavelength increases from left to right and bottom to top, and echelle order number increases from top to bottom. The spectral direction in each order is along the horizontal axis and the spatial direction is along the vertical axis. during each fall/winter observing season, and the number of spectra obtained. Additional

IGRINS data collected at McDonald Observatory were also used in this analysis and account for about 1/3 of the dataset. The targets were observed by nodding between the AB positions of the slit in either AB pairs or ABBA quads, in order to sample and remove the sky background. The SNR of all observations was typically ≥100. Standard A0 telluric stars at similar airmasses to the target observations were also observed. The A0 star observations were used to characterize the instrumental profile and to remove the telluric absorption features that are prevalent at near-infrared wavelengths, prior to the pre-main sequence star characterization analysis; conversely, the RV analysis relies on spectra that include the telluric absorption features.

The data were reduced with the standard IGRINS Python pipeline (v2.2.0 alpha 1) (Lee et al. 2017). The pipeline implements bad pixel correction, flat-fielding, sky background subtraction, spectral extraction, spectral distortion correction, and wavelength calibration.

Telluric correction is also optionally implemented by fitting a Gaussian to the A0 telluric

64 Table 2.3. Observations Summary

Observing Season N Nights N Spectra

2016-2017 12.50 140 2017-2018 19.00 254 2018-2019 8.25 211

Note. — Summary of IGRINS/LDT observations of the 70 pre-main sequence stars in the sample. Each observing season refers to the fall/winter of the specified year.

star H lines. The wavelength solution is determined by fitting night sky OH emission lines during a first pass, and telluric H2O absorption lines in the A0 telluric star during a second pass. Wavelength solutions are fit for each order to produce a distortion function, which describes the amount of shift in the spectral direction as a function of the slit position (in pixels). Essentially, this indicates where the reference wavelength solution is relative to the center of the slit. Figure 2.7 shows an example of the wavelength calibration. The final reduced output is a spectrum separated into 43 orders over the H and K bands. Additional details of the pipeline can be found in Sim et al. (2014).

An example of the fully reduced IGRINS spectra of the WTTS, V1075 Tau, is shown in Figure 2.8. The spectra were taken at different observing epochs, and a spectrum from each observing season is shown. This particular spectral order contains the region in the

K band used in the RV analysis, centered at ∼2.3µm. This region contains narrow and deep absorption lines caused by CO in the stellar photosphere, as well as numerous telluric features primarily caused by CH4 in the Earth’s atmosphere.

65 Figure 2.7 An example of IGRINS echellograms before (top) and after (middle) wavelength calibration and distortion correction. The wavelength solution is determined using night sky OH emission lines and H2O absorption lines in the A0 telluric star. The bottom panel shows the resulting spectrum.

66 2.3 Prior Optical Spectroscopy Survey

The infrared RV survey discussed in Section 2.2 builds on an optical RV survey that began in 2004. The objective was to observe pre-main sequence stars and look for RV signatures of short-period brown dwarfs and giant planets (Huerta et al. 2008; Prato et al. 2008;

Mahmud et al. 2011; Crockett et al. 2011, 2012; Johns-Krull et al. 2016). The following gives an overview of this prior survey, which forms the basis for the current infrared survey.

All the analysis and results presented in this subsection were completed by current or past collaborators who contributed to the initial young star RV survey prior to my joining the collaboration.

The visible light RV monitoring program (Huerta et al. 2008; Prato et al. 2008; Mahmud et al. 2011; Crockett et al. 2011, 2012; Johns-Krull et al. 2016) was conducted primarily with the Robert G. Tull Coud´eSpectrometer (Tull et al. 1995) on the 2.7-m Harlan J.

Smith Telescope, as well as the Sandiford Cassegrain Spectrograph (McCarthy et al. 1993) on the 2.1-m Otto Struve Telescope, and the Cassegrain echelle spectrograph on the 4-m

Kitt Peak National Observatory Mayall Telescope. The infrared follow-up was primarily conducted using the high-resolution Cassegrain-mounted echelle spectrograph, CSHELL

(Tokunaga et al. 1990; Greene et al. 1993) on the NASA Infrared Telescope Facility. This survey covered the ∼620–710 nm and ∼2.3 µm wavelength bands. The observations were typically obtained over a roughly week-long observing window in order to closely match both a possible companion’s orbital period and a typical pre-main seqeunce star’s rotation period.

The optical RVs were estimated using a cross-correlation technique over several spectral orders. The target star itself was used as the template for the cross-correlation to avoid the potential for spectral-type mismatching. RVs were therefore measured relative to a single observing epoch. RV variability, potentially signaling a companion detection, was then identified. In addition to the optical RVs, this survey also used simulated spectra of starspots

(Huerta et al. 2008), bisector analysis (a technique used to quantify the asymmetry of a

67 Figure 2.8 IGRINS spectra of the WTTS, V1075 Tau, from each observing season that IGRINS was at LDT. The strong absorption features primarily caused by CO in the stellar photosphere and CH4 in the Earth’s atmosphere are ideal for accurate RV analysis. The stellar absorption lines are roughly evenly-spaced throughout the spectra (the two isolated absorption lines at ∼2.312µm and ∼2.315µm are caused by stellar CO). The telluric lines are also present throughout the region with heavy absorption features between ∼2.316- 2.320µm. The follow-up observing season in 2018-2019 had lower resolution due to loose fasteners on the K band detector mount (bottom spectrum).

68 specific absorption or emission line from a spectrum’s cross-correlation function) (Huerta et al. 2008; Prato et al. 2008; Mahmud et al. 2011), and infrared spectra to determine if an

RV signal was caused by a substellar companion or starspots on the rotating surface of the star (Prato et al. 2008; Mahmud et al. 2011; Crockett et al. 2011, 2012; Johns-Krull et al.

2016).

Huerta et al. (2008) investigated the significant starspot-induced RV variation in the spectra of the WTTS, LkCa 19. They used simulations of spectra corresponding to different starspot scenarios to show that the observed RV variability was best modeled by the presence of a single large starspot on the star’s surface. They combined this method with line bisector analysis to test whether starspots could be the cause of the RV variation. A correlation between the RV variations and the changes in the line bisector slope indicates starspots as the cause of the RV variation. An example of their line bisector analysis is shown in Figure

2.9. The result of this analysis indicated modulation of the RVs at a period matching that of the stellar rotation period, derived from photometry. This analysis strongly suggests that the RV variability is caused by a single large starspot, rotating at the rotation period of the star, rather than by a companion’s orbital motion.

Prato et al. (2008) also used line bisector analysis of optical spectra to investigate RV variation in three young stars: DN Tau, V836 Tau, and V827 Tau. The V827 Tau dataset showed a clear correlation between the RV modulation and the bisector span, and therefore was determined to have significant starspots rather than a substellar companion. However,

DN Tau and V836 Tau did not show a clear correlation in the line bisector analysis. Desort et al. (2007) have shown that the RV and bisector variations caused by starspots can result in a lack of correlation in the line bisector analysis if the v sin i is smaller than the spectrograph resolution (as is the case for these two targets). Therefore, while line bisector analysis can prove activity as the cause of RV variation, it cannot prove the existence of a planetary companion.

Similar analysis was conducted by Mahmud et al. (2011) to probe the brown dwarf desert around pre-main sequence stars. Cross-correlation optical RVs combined with bisector

69 Figure 2.9 An example of line bisector analysis from Huerta et al. (2008). The RVs were measured by cross-correlating a single observed spectrum with the other spectra over multiple spectral orders, then measuring the shift of the cross correlation function. Asymmetry in this function is measured by the line bisector slope. A clear correlation between the bisector span and RVs is observed in the case that starspots are the source of the RV variation.

70 analysis strongly suggested that the RV variability of Hubble I 4, was spot-induced. The periodic signal detected in the optical spectra at 1.5 days matched the reported stellar rotation period (Norton et al. 2007), and provided further evidence that the RV variation was caused by starspots. Crockett et al. (2011, 2012) presented additional discussion of optical spectroscopic observations using similar techniques.

This initial survey resulted in the discovery of a hot Jupiter candidate around the 2

Myr old CTTS, CI Tau (Johns-Krull et al. 2016). The data were collected over 10 years at 5 different facilities (using both optical and infrared instruments). Periodogram analysis of the RVs indicated a likely planet orbital period of 9 days, and the RVs phased to that period demonstrate an RV variability amplitude indicating a planet mass of 11 MJ. Figure 2.10 shows the periodogram analysis and RV phase curve indicating the young hot Jupiter candidate discovery. One of the objectives of the current infrared RV survey is to conduct an intensive observing campaign and implement additional analysis methods to confirm this planet candidate (Section 4.2.1).

The prior optical RV survey also made use of infrared spectroscopy to support the results from the bisector analysis and visible light RV analysis. Prato et al. (2008) resolved the issue of uncorrelated RVs and bisector spans for two targets, DN Tau and V836 Tau, by comparing the RV amplitudes in infrared and optical light. They determined that the amplitudes were mismatched, and therefore the variation was not likely to be caused by a companion. Crockett et al. (2011, 2012) ruled out potential candidate hosts based on infrared data and identified CI Tau as an object for RV follow-up. Johns-Krull et al. (2016) investigated the optical and infrared RV measurements and determined that the data at both wavelengths phase coherently with equal amplitudes modulated at the 9-day orbital period. The ratio of the optical to infrared RV amplitudes was 0.64 ± 0.26, which was less than that expected for activity-induced signals, and therefore provided evidence for the planetary nature of this signal.

Several surveys have not succeeded at finding young exoplanets, in large part due to small sample sizes and intrinsic RV variability caused by the high levels of stellar activity observed

71 Figure 2.10 Detection of a young hot Jupiter candidate, CI Tau b, an 11 MJ planet in a 9-day orbital period around its 2 Myr old host star from Johns-Krull et al. (2016). The periodogram analysis (top) illustrates the 9-day orbital period detected in the dataset, with a false alarm probability of 10−4. 72 in pre-main sequence stars (Paulson et al. 2004; Paulson & Yelda 2006). Some surveys have successfully used optical spectroscopy to identify young planetary companions or candidates around WTTSs (Quinn et al. 2014; Donati et al. 2016; Yu et al. 2017). However, Prato et al. (2008), Mahmud et al. (2011), Crockett et al. (2011, 2012), and Johns-Krull et al.

(2016) made significant progress towards this goal by including follow-up infrared RVs to confirm or refute their findings from the optical data. Infrared RVs are a powerful tool for exoplanet detection, particularly when coupled with optical RVs and knowledge of the star’s rotation period, and can be used to search for young substellar companions around

CTTSs.

2.4 Optical Photometric Survey

In addition to the infrared spectra, this survey includes many epochs of time domain photometry for the targets in the sample. Photometry can be used to investigate stellar activity and determine stellar rotation periods. Near-simultaneous photometry can also be used in combination with spectroscopy to look for concurrent variability in both the photometric and RV datasets (Huerta et al. 2008). The ground-based photometry was collected and analyzed by collaborator Brian Skiff, and the space-based photometry was analyzed by collaborator Lauren Biddle.

The magnitudes of stars are typically measured relative to a comparison star of known brightness. The photometric observations are then investigated for periodicity using periodogram analysis and phased to the identified stellar rotation period. This lightcurve indicates the rotation period of the star, which can be used to determine the periodicity expected from starspots on the surface of young stars. Knowledge of the stellar rotation period can allow us to rule out a planetary source for RV variation modulated at that period

(although RV variation at the stellar rotation period could also be caused by a tidally-locked planet on a short orbit (Marcy et al. 1997)).

73 Using photometry to determine the stellar rotation period of young stars presents several challenges because of photometric variability (Bouvier et al. 1995; Preibisch & Smith 1997;

Stelzer et al. 2003; Stauffer et al. 2014). Instabilities in the inner disk around CTTSs cause accretion of disk material onto the stellar photosphere. This phenomenon can produce transient hot spots that cause variation in the measured brightness, leading to a chaotic lightcurve with no clear periodicity. Variable extinction, large starspots, rapid rotation, and stellar flares can also cause photometric variability, all of which vary on differing timescales

(Stauffer et al. 2014). Even with a high fraction of starspots, photometric modulation may not be identified if spots are located primarily in the circumpolar regions of the star, or are small and isotropic (Gully-Santiago et al. 2017). Long photometric time series can help mitigate these effects to accurately determine periodicity caused by stellar rotation (Messina et al. 2017).

Ground-based visible light photometry for many of the targets was obtained and supplemented with stellar rotation periods from the literature and K2 photometry. The ground-based photometry comes from seasonal V-band monitoring of the sample stars using the Lowell 0.7-m robotic telescope. The stars have been generally observed about ten nights per month from September through March each year since 2012, using CCD differential aperture photometry. Supplemental K2 long-cadence time-series photometry observed during Campaign 13 between 2017 March 8 and 2017 May 27 UTC was also used. The data were reduced using the K2SC pipeline (Aigrain et al. 2015, 2016) to remove instrumental systematics in the data while preserving astrophysical variability of the host star. Figure 2.11 shows an example K2 lightcurve for the pre-main sequence star, GM Tau,

and the periodogram analysis identifying the stellar rotation period of ∼2.7 days.

A subset of the sample targets were also observed by K2 and were prioritized as part of the planet search. The photometry of systems with identified companions could then be used to identify signatures related to star-planet-disk interactions (an effort led by collaborator,

Lauren Biddle; see Biddle et al. 2018 for details). The overlapping targets are AA Tau, DM

Tau, GI Tau, GM Tau (Figure 2.11), IQ Tau, and LkCa 15 systems.

74 1.0

1.5 0.8

0.6 2.667 1.0

Power 0.4

Normalized Flux 0.5 0.2

0.0 7820 7830 7840 7850 7860 7870 7880 7890 7900 0 5 10 15 20 BJD - 2450000 Period (d)

Figure 2.11 An example lightcuve from collaborator Lauren Biddle for the pre-main sequence star, GM Tau (left). The stellar flux was measured from K2 photometric observations. The stellar rotation period of ∼2.7 days is estimated based on the power spectrum from periodogram analysis (right).

75 References

Abramson, G. 2018, Research Notes of the AAS, 2, 150

Aigrain, S., Hodgkin, S. T., Irwin, M. J., Lewis, J. R., & Roberts, S. J. 2015, MNRAS, 447,

2880

Aigrain, S., Parviainen, H., & Pope, B. J. S. 2016, MNRAS, 459, 2408

Akeson, R. L., Jensen, E. L. N., Carpenter, J., Ricci, L., Laos, S., Nogueira, N. F., &

Suen-Lewis, E. M. 2019, ApJ, 872, 158

Andrews, S. M. & Williams, J. P. 2005, ApJ, 631, 1134

Bailey, John I., I., White, R. J., Blake, C. H., Charbonneau, D., Barman, T. S., Tanner,

A. M., & Torres, G. 2012, ApJ, 749, 16

Basri, G. & Batalha, C. 1990, ApJ, 363, 654

Bean, J. L., Seifahrt, A., Hartman, H., Nilsson, H., Wiedemann, G., Reiners, A., Dreizler,

S., & Henry, T. J. 2010, ApJ, 713, 410

Beichman, C. A., Boden, A., Ghez, A., Hartmann, L. W., Hillenbrand, L., Lunine,

J. I., Simon, M. J., Stauffer, J. R., & Velusamy, T. 2002, in Science with the Space

Interferometry Mission, ed. S. Unwin & S. Turyshev, 1

Benatti, S. 2018, Geosciences, 8, 289

Biddle, L. I., Johns-Krull, C. M., Llama, J., Prato, L., & Skiff, B. A. 2018, ApJ, 853, L34

76 Blake, C. H., Charbonneau, D., & White, R. J. 2010, ApJ, 723, 684

Bouvier, J., Covino, E., Kovo, O., Martin, E. L., Matthews, J. M., Terranegra, L., & Beck,

S. C. 1995, A&A, 299, 89

Bouy, H., Bertin, E., Sarro, L. M., Barrado, D., Moraux, E., Bouvier, J., Cuillandre,

J. C., Berihuete, A., Olivares, J., & Beletsky, Y. 2015, VizieR Online Data Catalog,

J/A+A/577/A148

Brice˜no,C., Luhman, K. L., Hartmann, L., Stauffer, J. R., & Kirkpatrick, J. D. 2002, ApJ,

580, 317

Carleo, I., Benatti, S., Lanza, A. F., Gratton, R., Claudi, R., Desidera, S., Mace, G. N.,

Messina, S., Sanna, N., Sissa, E., Ghedina, A., Ghinassi, F., Guerra, J., Harutyunyan,

A., Micela, G., Molinari, E., Oliva, E., Tozzi, A., Baffa, C., Baruffolo, A., Bignamini, A.,

Buchschacher, N., Cecconi, M., Cosentino, R., Endl, M., Falcini, G., Fantinel, D., Fini,

L., Fugazza, D., Galli, A., Giani, E., Gonz´alez,C., Gonz´alez-Alvarez,´ E., Gonz´alez,M.,

Hernandez, N., Hernandez Diaz, M., Iuzzolino, M., Kaplan, K. F., Kidder, B. T., Lodi,

M., Malavolta, L., Maldonado, J., Origlia, L., Perez Ventura, H., Puglisi, A., Rainer, M.,

Riverol, L., Riverol, C., San Juan, J., Scuderi, S., Seemann, U., Sokal, K. R., Sozzetti,

A., & Sozzi, M. 2018, A&A, 613, A50

Crockett, C. J., Mahmud, N. I., Prato, L., Johns-Krull, C. M., Jaffe, D. T., & Beichman,

C. A. 2011, ApJ, 735, 78

Crockett, C. J., Mahmud, N. I., Prato, L., Johns-Krull, C. M., Jaffe, D. T., Hartigan, P. M.,

& Beichman, C. A. 2012, ApJ, 761, 164

Daemgen, S., Bonavita, M., Jayawardhana, R., Lafreni`ere,D., & Janson, M. 2015, ApJ,

799, 155

Desort, M., Lagrange, A. M., Galland, F., Udry, S., & Mayor, M. 2007, A&A, 473, 983

77 Donati, J. F., Moutou, C., Malo, L., Baruteau, C., Yu, L., H´ebrard,E., Hussain, G.,

Alencar, S., M´enard,F., Bouvier, J., Petit, P., Takami, M., Doyon, R., & Cameron,

A. C. 2016, Nature, 534, 662

Fedele, D., van den Ancker, M. E., Henning, T., Jayawardhana, R., & Oliveira, J. M. 2010,

A&A, 510, A72

Figueira, P., Pepe, F., Lovis, C., & Mayor, M. 2010, A&A, 515, A106

Fischer, D. A., Laughlin, G., Butler, P., Marcy, G., Johnson, J., Henry, G., Valenti, J.,

Vogt, S., Ammons, M., Robinson, S., Spear, G., Strader, J., Driscoll, P., Fuller, A.,

Johnson, T., Manrao, E., McCarthy, C., Mu˜noz,M., Tah, K. L., Wright, J., Ida, S.,

Sato, B., Toyota, E., & Minniti, D. 2005, ApJ, 620, 481

Fontanive, C., Rice, K., Bonavita, M., Lopez, E., Muˇzi´c,, K., & Biller, B. 2019, MNRAS,

485, 4967

Ghez, A. M., Neugebauer, G., & Matthews, K. 1993, AJ, 106, 2005

Greene, T. P., Tokunaga, A. T., Toomey, D. W., & Carr, J. B. 1993, Society of Photo-

Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 1946, CSHELL: a high

spectral resolution 1-5 um cryogenic echelle spectrograph for the IRTF, ed. A. M. Fowler,

313–324

Gully-Santiago, M. A., Herczeg, G. J., Czekala, I., Somers, G., Grankin, K., Covey, K. R.,

Donati, J. F., Alencar, S. H. P., Hussain, G. A. J., Shappee, B. J., Mace, G. N., Lee,

J.-J., Holoien, T. W. S., Jose, J., & Liu, C.-F. 2017, ApJ, 836, 200

Hall, D. S. & Henry, G. W. 1994, International Amateur-Professional Photoelectric

Photometry Communications, 55, 51

Hartigan, P., Hartmann, L., Kenyon, S., Hewett, R., & Stauffer, J. 1989, ApJS, 70, 899

Hartigan, P., Kenyon, S. J., Hartmann, L., Strom, S. E., Edwards, S., Welty, A. D., &

Stauffer, J. 1991, ApJ, 382, 617

78 Haywood, R. 2016, Stellar Activity as a Source of Radial-Velocity Variability, 13–44

Herbig, G. H. & Bell, K. R. 1988, Third Catalog of Emission-Line Stars of the Orion

Population : 3 : 1988

Herbst, W., Herbst, D. K., Grossman, E. J., & Weinstein, D. 1994, AJ, 108, 1906

Hern´an-Obispo, M., G´alvez-Ortiz, M. C., Anglada-Escud´e,G., Kane, S. R., Barnes, J. R.,

de Castro, E., & Cornide, M. 2010, A&A, 512, A45

Hu´elamo,N., Figueira, P., Bonfils, X., Santos, N. C., Pepe, F., Gillon, M., Azevedo, R.,

Barman, T., Fern´andez, M., di Folco, E., Guenther, E. W., Lovis, C., Melo, C. H. F.,

Queloz, D., & Udry, S. 2008, A&A, 489, L9

Huerta, M., Johns-Krull, C. M., Prato, L., Hartigan, P., & Jaffe, D. T. 2008, ApJ, 678, 472

Johns-Krull, C. M. 2007, ApJ, 664, 975

Johns-Krull, C. M., McLane, J. N., Prato, L., Crockett, C. J., Jaffe, D. T., Hartigan, P. M.,

Beichman, C. A., Mahmud, N. I., Chen, W., Skiff, B. A., Cauley, P. W., Jones, J. A., &

Mace, G. N. 2016, ApJ, 826, 206

Johnson, J. A., Aller, K. M., Howard, A. W., & Crepp, J. R. 2010, PASP, 122, 905

Kraus, A. L., Herczeg, G. J., Rizzuto, A. C., Mann, A. W., Slesnick, C. L., Carpenter,

J. M., Hillenbrand, L. A., & Mamajek, E. E. 2017, ApJ, 838, 150

Kraus, A. L. & Hillenbrand, L. A. 2009, ApJ, 704, 531

Kraus, A. L., Ireland, M. J., Hillenbrand, L. A., & Martinache, F. 2012, ApJ, 745, 19

Kraus, A. L., Ireland, M. J., Martinache, F., & Hillenbrand, L. A. 2011, ApJ, 731, 8

Lee, J.-J., Gullikson, K., & Kaplan, K. 2017, igrins/plp 2.2.0

Lockwood, G. W., Skiff, B. A., Henry, G. W., Henry, S., Radick, R. R., Baliunas, S. L.,

Donahue, R. A., & Soon, W. 2007, ApJS, 171, 260

79 Luhman, K. L. 2018, AJ, 156, 271

Mace, G., Sokal, K., Lee, J.-J., Oh, H., Park, C., Lee, H., Good, J., MacQueen, P., Oh,

J. S., Kaplan, K., Kidder, B., Chun, M.-Y., Yuk, I.-S., Jeong, U., Pak, S., Kim, K.-

M., Nah, J., Lee, S., Yu, Y.-S., Hwang, N., Park, B.-G., Kim, H., Chinn, B., Peck, A.,

Diaz, R., Rutten, R., Prato, L., Jacoby, G., Cornelius, F., Hardesty, B., DeGroff, W.,

Dunham, E., Levine, S., Nofi, L., Lopez-Valdivia, R., Weinberger, A. J., & Jaffe, D. T.

2018, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series,

Vol. 10702, Proc. SPIE, 107020Q

Mahmud, N. I., Crockett, C. J., Johns-Krull, C. M., Prato, L., Hartigan, P. M., Jaffe, D. T.,

& Beichman, C. A. 2011, ApJ, 736, 123

Manara, C. F., Morbidelli, A., & Guillot, T. 2018, A&A, 618, L3

Marcy, G. W., Butler, R. P., Williams, E., Bildsten, L., Graham, J. R., Ghez, A. M., &

Jernigan, J. G. 1997, ApJ, 481, 926

Mayor, M., Marmier, M., Lovis, C., Udry, S., S´egransan,D., Pepe, F., Benz, W., Bertaux,

J. ., Bouchy, F., Dumusque, X., Lo Curto, G., Mordasini, C., Queloz, D., & Santos, N. C.

2011, ArXiv e-prints

McCabe, C., Ghez, A. M., Prato, L., Duchˆene,G., Fisher, R. S., & Telesco, C. 2006, ApJ,

636, 932

McCarthy, J. K., Sandiford, B. A., Boyd, D., & Booth, J. 1993, PASP, 105, 881

Messina, S., Parihar, P., & Distefano, E. 2017, MNRAS, 468, 931

Meynet, G., Mermilliod, J. C., & Maeder, A. 1993, A&AS, 98, 477

Monin, J. L., Clarke, C. J., Prato, L., & McCabe, C. 2007, in Protostars and Planets V,

ed. B. Reipurth, D. Jewitt, & K. Keil, 395

80 Monin, J. L., Guieu, S., Pinte, C., Rebull, L., Goldsmith, P., Fukagawa, M., M´enard, F.,

Padgett, D., Stappelfeld, K., McCabe, C., Carey, S., Noriega-Crespo, A., Brooke, T.,

Huard, T., Terebey, S., Hillenbrand, L., & Guedel, M. 2010, A&A, 515, A91

Najita, J. R. & Kenyon, S. J. 2014, MNRAS, 445, 3315

Ngo, H., Knutson, H. A., Hinkley, S., Bryan, M., Crepp, J. R., Batygin, K., Crossfield, I.,

Hansen, B., Howard, A. W., Johnson, J. A., Mawet, D., Morton, T. D., Muirhead, P. S.,

& Wang, J. 2016, ApJ, 827, 8

Norton, A. J., Wheatley, P. J., West, R. G., Haswell, C. A., Street, R. A., Collier Cameron,

A., Christian, D. J., Clarkson, W. I., Enoch, B., Gallaway, M., Hellier, C., Horne, K.,

Irwin, J., Kane, S. R., Lister, T. A., Nicholas, J. P., Parley, N., Pollacco, D., Ryans, R.,

Skillen, I., & Wilson, D. M. 2007, A&A, 467, 785

Park, C., Jaffe, D. T., Yuk, I.-S., Chun, M.-Y., Pak, S., Kim, K.-M., Pavel, M., Lee, H., Oh,

H., Jeong, U., Sim, C. K., Lee, H.-I., Nguyen Le, H. A., Strubhar, J., Gully-Santiago,

M., Oh, J. S., Cha, S.-M., Moon, B., Park, K., Brooks, C., Ko, K., Han, J.-Y., Nah,

J., Hill, P. C., Lee, S., Barnes, S., Yu, Y. S., Kaplan, K., Mace, G., Kim, H., Lee, J.-J.,

Hwang, N., & Park, B.-G. 2014, in Proc. SPIE, Vol. 9147, Ground-based and Airborne

Instrumentation for Astronomy V, 91471D

Paulson, D. B., Cochran, W. D., & Hatzes, A. P. 2004, AJ, 127, 3579

Paulson, D. B. & Yelda, S. 2006, PASP, 118, 706

P´ericaud,J., Di Folco, E., Dutrey, A., Guilloteau, S., & Pi´etu,V. 2017, A&A, 600, A62

Perryman, M. A. C., Brown, A. G. A., Lebreton, Y., Gomez, A., Turon, C., Cayrel de

Strobel, G., Mermilliod, J. C., Robichon, N., Kovalevsky, J., & Crifo, F. 1998, A&A,

331, 81

Prato, L., Huerta, M., Johns-Krull, C. M., Mahmud, N., Jaffe, D. T., & Hartigan, P. 2008,

ApJ, 687, L103

81 Preibisch, T. & Smith, M. D. 1997, A&A, 322, 825

Quinn, S. N., White, R. J., Latham, D. W., Buchhave, L. A., Torres, G., Stefanik, R. P.,

Berlind, P., Bieryla, A., Calkins, M. C., Esquerdo, G. A., F˝ur´esz,G., Geary, J. C., &

Szentgyorgyi, A. H. 2014, ApJ, 787, 27

Radick, R. R., Lockwood, G. W., Skiff, B. A., & Baliunas, S. L. 1998, ApJS, 118, 239

Rajpaul, V., Aigrain, S., & Roberts, S. 2016, MNRAS, 456, L6

Reiners, A., Bean, J. L., Huber, K. F., Dreizler, S., Seifahrt, A., & Czesla, S. 2010, ApJ,

710, 432

Saar, S. H. & Donahue, R. A. 1997, ApJ, 485, 319

Schaefer, G. H., Prato, L., Simon, M., & Zavala, R. T. 2012, ApJ, 756, 120

Setiawan, J., Henning, T., Launhardt, R., M¨uller,A., Weise, P., & K¨urster, M. 2008,

Nature, 451, 38

Sim, C. K., Le, H. A. N., Pak, S., Lee, H.-I., Kang, W., Chun, M.-Y., Jeong, U., Yuk, I.-S.,

Kim, K.-M., Park, C., Pavel, M. D., & Jaffe, D. T. 2014, Advances in Space Research,

53, 1647

Simon, M., Ghez, A. M., Leinert, C., Cassar, L., Chen, W. P., Howell, R. R., Jameson,

R. F., Matthews, K., Neugebauer, G., & Richichi, A. 1995, ApJ, 443, 625

Spruit, H. C. 1976, Sol. Phys., 50, 269

Stauffer, J., Cody, A. M., Baglin, A., Alencar, S., Rebull, L., Hillenbrand, L. A., Venuti,

L., Turner, N. J., Carpenter, J., Plavchan, P., Findeisen, K., Carey, S., Terebey, S.,

Morales-Calder´on,M., Bouvier, J., Micela, G., Flaccomio, E., Song, I., Gutermuth, R.,

Hartmann, L., Calvet, N., Whitney, B., Barrado, D., Vrba, F. J., Covey, K., Herbst, W.,

Furesz, G., Aigrain, S., & Favata, F. 2014, AJ, 147, 83

82 Stelzer, B., Fern´andez, M., Costa, V. M., Gameiro, J. F., Grankin, K., Henden, A.,

Guenther, E., Mohanty, S., Flaccomio, E., Burwitz, V., Jayawardhana, R., Predehl,

P., & Durisen, R. H. 2003, A&A, 411, 517

Tokunaga, A. T., Toomey, D. W., Carr, J., Hall, D. N. B., & Epps, H. W. 1990, Society

of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 1235, Design

for a 1–5-um cryogenic echelle spectrograph for the NASA IRTF, ed. D. L. Crawford,

131–143

Tull, R. G., MacQueen, P. J., Sneden, C., & Lambert, D. L. 1995, PASP, 107, 251

Udry, S. & Santos, N. C. 2007, ARA&A, 45, 397

Valenti, J. A., Basri, G., & Johns, C. M. 1993, AJ, 106, 2024

Yan, Q.-Z., Zhang, B., Xu, Y., Guo, S., Macquart, J.-P., Tang, Z.-H., & Walsh, A. J. 2019,

Astronomy and Astrophysics, 624, A6

Yu, L., Donati, J.-F., H´ebrard,E. M., Moutou, C., Malo, L., Grankin, K., Hussain, G.,

Collier Cameron, A., Vidotto, A. A., Baruteau, C., Alencar, S. H. P., Bouvier, J., Petit,

P., Takami, M., Herczeg, G., Gregory, S. G., Jardine, M., Morin, J., M´enard,F., &

Matysse Collaboration. 2017, MNRAS

Yuk, I.-S., Jaffe, D. T., Barnes, S., Chun, M.-Y., Park, C., Lee, S., Lee, H., Wang, W., Park,

K.-J., Pak, S., Strubhar, J., Deen, C., Oh, H., Seo, H., Pyo, T.-S., Park, W.-K., Lacy, J.,

Goertz, J., Rand, J., & Gully-Santiago, M. 2010, in Proc. SPIE, Vol. 7735, Ground-based

and Airborne Instrumentation for Astronomy III, 77351M

Zhang, Z., Liu, M. C., Best, W. M. J., Magnier, E. A., Aller, K. M., Chambers, K. C.,

Draper, P. W., Flewelling, H., Hodapp, K. W., Kaiser, N., Kudritzki, R. P., Metcalfe, N.,

Wainscoat, R. J., & Waters, C. 2018, ApJ, 858, 41

Zucker, S. & Mazeh, T. 2002, ApJ, 568, L113

83 Chapter 3

Stellar Properties of Pre-Main Sequence Stars

This chapter originally appeared as Nofi et al. (submitted 2020), with co-authors

Christopher M. Johns–Krull, Ricardo L´opez–Valdivia, Lauren Biddle, Adolfo S. Carvalho,

Daniel Huber, Daniel Jaffe, Joe Llama, Gregory Mace, Lisa Prato, Brian Skiff, Kimberly

R. Sokal, Kendall Sullivan, and Jamie Tayar. This chapter presents results from the spectroscopic survey characterizing the stellar parameters of pre-main sequence stars (projected rotational velocity, effective temperature, rotation period, and limit on the radius). Collaborator Ricardo L´opez–Valdivia analyzed and provided the effective temperatures, Brian Skiff and Lauren Biddle analyzed and provided the stellar rotation periods, and Chris Johns–Krull and Adolfo Carvalho analyzed and provided optical projected rotational velocities that are compared to the infrared values. I conducted and analyzed the infrared projected rotational velocity survey that is the main focus of this chapter.

Abstract

The projected stellar rotational velocity (v sin i) is critical for an understanding of processes related to the evolution of angular momentum in pre-main sequence stars. Presented here are v sin i measurements of high-resolution infrared and optical spectroscopy for 70 pre-main sequence stars in the Taurus-Auriga star-forming region, in addition to effective temperatures measured from line-depth ratios, and stellar rotation periods determined from

84 optical photometry. The stars in the sample that show evidence of residing in circumstellar disks or multiple systems were identified from the literature. The comparison of infrared v sin i measurements calculated using two techniques shows a residual scatter of ∼1.8 km

s−1, defining a typical error floor for the v sin i of pre-main sequence stars from infrared spectra. A comparison of the v sin i distributions of stars with and without companions shows that binaries/multiples typically have a higher measured v sin i, which may be caused by contamination by companion lines, shorter disk lifetimes in binary systems, or tidal interactions in hierarchical triples. A comparison of optical and infrared v sin i values shows no significant difference regardless of whether the star has a disk or not, indicating that CO contamination from the disk does not impact v sin i measurements above the typical ∼1.8 km s−1 error floor of the measurements. Finally, no correlation between the v sin i, presence of a disk, and H-R diagram position was observed, which indicates a complex interplay between stellar rotation and evolution of pre-main sequence stars.

3.1 Introduction

The projected stellar rotational velocity (v sin i, where i is the inclination) is an important observable that provides information about the angular momentum evolution of a star.

Stellar rotation has been linked to dynamo-driven magnetic activity (Hartmann & Noyes

1987; Gameiro & Lago 1993; Bouvier 2013), mass loss through stellar winds (Tout &

Pringle 1992; Matt & Pudritz 2005a; Johnstone 2017a), interactions with circumstellar disks (Attridge & Herbst 1992; Mathieu 2003), disk lifetimes (Bouvier 1997), stellar internal structure, differential rotation, internal transport of angular momentum (Wolff et al. 2004;

Blackman & Thomas 2015), and stellar birth environments (Choi & Herbst 1996; Spruit

2018). Rotation produces a systematic lowering of the effective temperature of stars which can be as large as ∼300 K for very rapid rotation (Sills et al. 2000). Measurements of v sin i, as a supporting criterion of youth, have been used to aid in membership studies, confirming kinematic members within young open clusters (G´alvez-Ortiz et al. 2010). The v sin i can

85 inform phenomena related to the extended main sequence turn-off, which constrains cluster ages (Bastian et al. 2018; Cummings & Kalirai 2018; Sun et al. 2019). It is also an important parameter when considering interactions between two bodies, such as stellar binary interactions involving disk truncation and the resulting stellar spin-up (Stauffer et al. 2018), and stellar spin history related to planet formation, evolution, and star-planet interactions (Cohen et al. 2010; Bolmont et al. 2012; Bouvier & C´ebron2015; Benbakoura et al. 2019). Furthermore, stellar rotation-induced activity and winds cause mass loss of planetary atmospheres, and therefore play a significant role in planet habitability in main sequence systems (Kislyakova et al. 2013; Johnstone et al. 2015; Vidotto 2016; Johnstone

2017b).

The rotational velocities of pre-main sequence stars have been particularly significant in contributing to an understanding of the evolution of angular momentum in young stars.

Young clusters of different ages have been studied to understand the rotational history of low-mass stars. Early observations of pre-main sequence rotational velocities revealed that this population typically rotates more slowly than expected based on stellar evolution theory alone, at a small fraction of the star’s break-up velocity (Vogel & Kuhi 1981; Bouvier et al. 1986; Hartmann & Stauffer 1989; Bouvier 1990; Clarke & Bouvier 2000). To account for the observed low v sin i (∼15 km s−1) of this population, a strong braking mechanism was needed (Bouvier et al. 1993b). One proposed mechanism is magnetic disk-locking, in which the star experiences magnetic braking through interactions with the circumstellar disk, causing the young star to lose angular momentum (Camenzind 1990; Koenigl 1991;

Attridge & Herbst 1992; Tinker et al. 2002; Mathieu 2003). This mechanism is supported by several studies showing that the median v sin i is lower for those populations that show evidence of hosting disks (Bouvier et al. 1993a; Edwards et al. 1993; Choi & Herbst 1996;

Herbst et al. 2002; Sicilia-Aguilar et al. 2005; Dahm et al. 2012; Xiao et al. 2012). Therefore, the resulting distributions of rotational velocities and disk dispersal timescales can inform disk-coupling lifetimes (Edwards et al. 1993; Bouvier 1997; Sills et al. 2000; Tinker et al.

2002). Stars with long-lived disks become relatively slow rotators and stars with short-lived

86 disks become relatively fast rotators (Stauffer et al. 1997). An additional braking mechanism of young stars is a magnetically-coupled stellar wind, which results in a significant loss of angular momentum (Hartmann & Stauffer 1989; Tout & Pringle 1992; Bolmont et al. 2012).

Once disk dispersal takes place, the pre-main sequence star spins up as it contracts toward the zero-age main sequence (Shu et al. 1994; Matt & Pudritz 2005b; Bouvier 2013).

This chapter presents v sin i measurements derived from infrared spectroscopy for a sample of 70 pre-main sequence stars. Additionally, optically-measured v sin i estimates to compare to the infrared v sin i values, effective temperatures (Teff , also from infrared spectroscopy), stellar rotation periods (Prot) from optical photometry, and consequently, lower limits on stellar radii (R sin i), are also presented. The sample consists of CTTSs and WTTSs, as identified in previously published studies that looked for evidence of an accreting circumstellar disk. The sample also includes both single stars or stars in multiple systems, also classified from the literature. Possible trends and correlations between stellar parameters are investigated; in particular, I compare the v sin i distributions of stars with

and without disks to test stellar evolution theory and infer the influence of a circumstellar

disk on stellar rotation. See Chapter 2 for a discussion of the sample selection, observations,

and analysis of the infrared and optical spectroscopy, and photometry.

3.2 Analysis

3.2.1 Stellar Effective Temperatures

The Teff analysis and results were provided by collaborator Ricardo L´opez–Valdivia using

line-depth ratios. L´opez-Valdivia et al. (2019) determined the Teff for 254 K and M main sequence stars from H band IGRINS spectra. They provided two linear relations between

Teff and the line-depth ratios (LDR) of Fe I, OH, and Al I absorption lines. These relations

offer a simple and accurate measure of Teff in cool stars, and can be used for young stellar objects (YSOs) because the LDRs are insensitive to veiling. L´opez-Valdivia et al. (2019) tested one of their relations on 12 members of the nearby (<60 pc; Gaia Collaboration 2018)

87 and young (∼7-10 Myr; Sokal et al. 2018) TW Hydrae Association (Kastner et al. 1997).

They found hotter temperatures (varying by ∼140 K) for stars with Teff between 3200 and 3800 K compared with previous determinations. This discrepancy may be the result of

differences in surface gravity between YSOs and the main sequence stars with which the

L´opez-Valdivia et al. (2019) approach was calibrated. The Teff - LDR relations are valid between ∼3100 and 4100 K. Most of the sample stars have Teff within the applicability limits of this approach (see Table 3.2).

The line-depths of the Fe I (λ ∼1.56216 µm) line and the OH (λ ∼1.56270 µm) doublet were measured in the same way as L´opez-Valdivia et al. (2019). The LDR(Fe/OH) was then computed and used in the linear equation T = 520 × LDR(Fe/OH) + 3230 K. For the

28 targets with Teff outside the limits of the relation, the value of Teff was extrapolated. A typical measurement error in the LDR analysis is ∼140 K, and a typical systematic error is

∼120 K. Therefore, a total uncertainty of ∼180 K was determined and a conservative error of 200 K was assigned for the whole sample.

3.2.2 Infrared v sin i

The infrared v sin i analysis used the IGRINS K band order with wavelength range 2.299-

2.319 µm. This region contains CO absorption lines, which are less sensitive to magnetic

fields or pressure broadening (Kesseli et al. 2018). This region also contains telluric absorption lines, which were removed using A0 star division. Magnetically-sensitive Ti

I lines in this region were masked to avoid measuring broadening of the lines from magnetic

fields. The final spectrum was continuum-normalized and trimmed to avoid inaccuracies caused by distortion at the ends of the spectral order. A sample of IGRINS spectra is shown in Figure 3.1. The range of wavelengths shown represents a subset of the region analyzed in this work. The figure demonstrates how the stellar spectral lines broaden with increasing v sin i for three targets with similar Teff . The infrared v sin i values were measured using the technique outlined in Hartmann et al. (1986) and Soderblom et al. (1989). The basic premise of this technique is that

88 Figure 3.1 IGRINS K band continuum-normalized spectra of three targets with a range of v sin i values. The wavelength region shown is a subset of the analysis region with CO absorption lines. Included are spectra of DI Tau (v sin i ∼10 km s−1 – blue), V1115 Tau (v sin i ∼20 km s−1 – orange), and V1075 Tau (v sin i ∼30 km s−1 – black). The stellar spectral lines become broader as the v sin i increases.

89 when a stellar spectrum with rotationally broadened lines is cross-correlated against an unbroadened, narrow-lined spectrum, the resulting width of the cross-correlation function

(CCF) can be used to measure the rotational broadening of the first spectrum. To measure the v sin i of the targets, I first calibrated a relation between the full width at half maximum

(FWHM) of the CCF and the v sin i. I then artificially broadened synthetic spectra over a range of v sin i from 1 to 20 km s−1 in steps of 1 km s−1, then up to a v sin i of 60 km s−1

in steps of 5 km s−1. This generated a calibration function that relates the FWHM of a

CCF to the v sin i. The process was repeated with stellar models of different temperatures

to establish calibration relations that matched the range of temperatures of the targets.

An unbroadened synthetic spectrum with a similar Teff to the target star was then chosen and cross-correlated with the target spectrum to measure the FWHM of the CCF, which

can then be related to the corresponding v sin i from the calibration relation. Figure 3.2 shows calibration relations for several synthetic spectra used in the analysis. While other phenomena can broaden spectral lines in addition to rotation, the FWHM of the CCF is a good indicator of v sin i for the sample because pre-main sequence stars have relatively low gravities in their atmospheres, which results in narrow photospheric absorption features in which rotational broadening dominates over pressure broadening (Dahm et al. 2012).

Typically, the FWHM of the CCF is measured from a baseline of zero. However, the stellar absorption lines in the region of interest are nearly evenly-spaced (see Figure 3.1), which creates a sinusoidal pattern in the CCF, causing the baseline of the CCF response to be negative. To accurately measure the FWHM, I instead measured from a baseline determined by the minimum of the CCF response. To account for variations due to noise, the average of the two minima on either side of the CCF peak was calculated (within

±100 pixels, which corresponds to about ±200 km s−1) as the baseline for the FWHM

measurement. This modification was made both when creating the calibration relations,

and measuring the FWHM of the CCF from the dataset to determine the v sin i.

While several studies have determined v sin i with this technique, most rely on observed spectral templates (Rhode et al. 2001; Wolff et al. 2004; Nordhagen et al. 2006; Nguyen

90 Figure 3.2 Relations between the FWHM of the CCF and v sin i for different stellar parameters: effective temperature (top), surface gravity (middle), and macroturbulence velocity (bottom). Using these relations, the v sin i was determined from a measurement of the FWHM of the CCF, which indicates the level of rotational broadening of the spectral lines. The relations do not vary significantly based on differences in the model parameters.

91 et al. 2009; Dahm et al. 2012). This work relies on synthetic spectra at infrared wavelengths to create a calibration relation and measure the v sin i of the targets. There is a good match between observational and theoretical spectra in the K band (Lyubchik et al.

2012). Rotational velocity standard templates may have significantly different physical properties than the target stars (Lyubchik et al. 2012) and are non-zero rotators, which leads to underestimated rotational velocities (Vogel & Kuhi 1981). Synthetic infrared spectra covering the region of interest were provided by collaborator, Chris Johns–Krull, and computed using the SYNTHMAG spectrum synthesis code (Piskunov 1999). Line data needed to compute the synthetic spectra were obtained from the Vienna Atomic Line

Database (Piskunov et al. 1995; Ryabchikova & Pakhomov 2015). The region of interest is dominated by lines of the CO molecule, and the line data in VALD for this molecule comes from Goorvitch (1994). The NextGen stellar atmosphere models (Allard & Hauschildt 1995) were used for the synthetic spectrum calculation. In all cases a microturbulent broadening of 1 km s−1 was assumed (D’Orazi et al. 2011), and the macroturbulent broadening

(discussed below) uses the radial-tangential formulation (Gray 2008) with the radial and tangential turbulent velocities equal. The synthetic spectra were convolved with a Gaussian to represent the standard resolution of the IGRINS spectra (R∼45,000) before they were used as a template to measure v sin i. The synthetic models were generated at varying Teff , surface gravity (log g), and macroturbulence velocity (vmac). Metallicity remained fixed and solar metallicity was assumed for the sample (D’Orazi et al. 2011). I conducted tests to investigate which of these stellar parameters are significant to the outcome of the measured v sin i. I explored a range of temperatures between 3100 K and 5100 K (corresponding to the range of Teff estimates of the targets) in steps of 200 K (the uncertainty in the Teff

−1 −1 measurements), a log g of 3.5 and 4.0, and a vmac of 2 km s and 3 km s . Figure 3.2 shows how the relation for determining v sin i based on the FWHM of the CCF changes with

Teff (a), log g (b) and by vmac (c). Varying the Teff , log g, and vmac did not significantly contribute to the total uncertainty within the range of parameter values of the targets. For

92 −1 these measurements, I assumed a vmac of 2 km s , and a log g of 3.5, since the sample contains cool, low-mass stars (Gray 2008).

To obtain the infrared v sin i measurements, each target was matched with a synthetic spectrum based on the measured Teff . The v sin i was then measured for each individual epoch and the average calculated to determine a final v sin i value for each target. A random internal uncertainty was determined by measuring the standard deviation of the mean, which on average resulted in an error of ∼0.7 km s−1.

Additionally, systematic errors due to different models and methodologies were

considered. The most likely source of systematic error is model choice mismatch (Hartmann

et al. 1986). While the uncertainty based on model choice can be significant (Lyubchik

et al. (2012) found that deviations of Teff , log g, and vmac lead to uncertainties in the determination of v sin i at the level of 1-2 km s−1), given the consistency between the

calibration relations based on alternate model choices (see Figure 3.2), I determined that

the systematic error is negligible in this analysis. I also characterized a systematic error

due to different methodologies by comparing the v sin i values to L´opez-Valdivia et al. (in

prep.), who measured the v sin i for an overlapping sub-sample of 51 targets. Instead of a

cross correlation technique, they used a Markov Chain Monte Carlo (MCMC) algorithm to

compare observations with MOOGStokes synthetic spectra (Deen 2013). They also included

additional spectral regions in their analysis. Both analyses relied on IGRINS data: this

work measured the v sin i for single spectra and averaged the result, while L´opez-Valdivia

et al. (in prep.) measured v sin i from a single averaged spectrum. Figure 3.3 shows a

comparison of the preliminary MCMC v sin i estimates and those from this work. There

is good agreement between the two results, and the standard deviation of the residuals

indicates a methodology uncertainty of ∼1.8 km s−1. This methodology error was added in

quadrature to the internal random error to determine a total uncertainty.

Rotational velocity measurements are limited by the SNR and spectral resolution of the

instrument. To determine this limit, I measured the v sin i of a radial velocity standard

star, GJ 281 (v sin i=1.0±0.9 km s−1; Smith 2015) which indicated a lower limit of ∼3-4

93 Figure 3.3 A comparison of v sin i measurements and residuals using IGRINS spectra and two different methods. L´opez-Valdivia et al. (in prep.) used an MCMC technique and MOOGStokes model spectra, while this work relies on a cross correlation technique and NextGen model spectra. The median offset is 1.0 km s−1 and the residual standard deviation is ∼1.8 km s−1, which is added in quadrature to the formal uncertainties.

94 km s−1. This value is consistent with detection thresholds reported for similar resolution spectra (Browning et al. 2010; Reiners et al. 2012; Davison et al. 2015; Kesseli et al. 2018).

Any v sin i estimate below this value is considered unreliable, however, I did not measure any v sin i values below this limit in the sample.

The broadening of the CCF also becomes difficult to accurately measure for fast rotators.

I estimated an upper limit on the v sin i for stars with high rotational velocities at which it is not possible to accurately measure the v sin i. This limit was established by cross- correlating an artificially broadened model, generated with noise corresponding to a SNR of

100, with the original unbroadened synthetic model for different values of v sin i. At ∼50 km s−1, a v sin i is recovered that is inaccurate at the level of the median random uncertainty measured for the dataset. Therefore, I established an upper limit on fast rotators of 50 km s−1.

3.2.3 Optical v sin i

The optical v sin i values were measured by collaborators Chris Johns–Krull and Adolfo

Carvalho using essentially the same technique described above with two exceptions. The

first was that the template used was the observed spectrum of HD 65277 instead of a synthetic spectrum. HD 65277 is classified as a K4 dwarf star (Houk & Swift 1999), which implies an effective temperature of 4620 K (Pecaut & Mamajek 2013). Valenti & Fischer

(2005) included HD 65277 in their spectroscopic analysis of 1040 F, G, and K dwarfs,

−1 finding a Teff =4741 K and a v sin i = 1.0 km s . The recent spectroscopic analysis of Soto

& Jenkins (2018) finds a stellar Teff = 4660 ± 14 K, along with a v sin i = 1.81 ± 0.08 km s−1. As these estimates of the rotational velocity are below the resolution of the optical data (∼ 2.5 km s−1), HD 65277 can serve as essentially a non-rotating template with a spectral type very similar to the sample.

The second difference is that the least squares deconvolution (LSD) technique (Donati et al. 1997) was employed to boost the signal-to-noise of the individual observations. This technique combines data from many spectral lines in a way similar to a cross-correlation

95 analysis. The basic idea is the assumption that the observed spectrum is the convolution between a spectrum composed of delta functions at the wavelength of each line and a single broadening function which takes into account all sources of line broadening including stellar rotation and the instrumental profile of the spectrograph. The strength of the different delta functions is directly proportional to the expected depth of the respective lines. The LSD code described in Chen & Johns-Krull (2013) was used to create the LSD profiles. The line list used (rest wavelengths and predicted depths) is based on the list constructed by Chen

& Johns-Krull (2013) for the analysis of BP Tau, a K7 CTTS, but trimmed down to 375 lines spanning the wavelength range from 528.8 nm to 886.6 nm. The LSD technique then deconvolves the observed spectrum with the line list to produce a high S/N Stokes I profile.

The FWHM of this LSD profile is measured to determine the v sin i of the star.

To construct the relationship between the FWHM of the LSD profile and the stellar v sin i the observed spectrum of HD 65277 was artificially broadened with a standard rotational broadening kernel (Gray 2008). The LSD profile of this rotationally broadened spectrum was then computed and the FWHM of the LSD profile was measured. This procedure was repeated for several values of v sin i to produce a calibration curve similar to Figure 3.2. For each pre-main sequence star, the LSD profile was computed and linearly interpolated on the calibration relationship to determine its v sin i. Most of the stars in the

sample have multiple observations, so this procedure was repeated for every observation.

The final value was determined from the mean of the different v sin i values. The random

uncertainty associated with the measurements was computed from the standard deviation

of the mean. For a few stars where there was only one observation, the random uncertainty

was assigned based on the standard deviation of other stars in the sample with similar

values for the v sin i. The systematic uncertainty in the analysis was estimated by using

two additional slowly rotating stars as templates, achieving agreement to typically better

than 0.7 km s−1. This was taken as the systematic uncertainty and added in quadrature to the random uncertainties. The v sin i from the optical spectra can be reliably measured

down to 4 km s−1 and up to 50 km s−1.

96 3.2.4 Stellar Rotation Periods

Ground-based photometry from the Lowell 0.7-m robotic telescope was obtained and analyzed by collaborator Brian Skiff using a Fourier-fitting routine (Harris et al. 1989). This technique used a phase-dispersion minimization routine to identify peaks in the periodogram relating to periodicity in the photometric data. A search for periodicities between 0.2 and

100 days in 0.01 day increments was implemented. A single, persistent period was chosen based on the results of the minimization routine. The period was assigned an uncertainty of ∼0.1 days, based on season-to-season scatter in the fitted lightcurves. Additional details of the Lowell photometry analysis can be found in Johns-Krull et al. (2016).

The K2 data from Campaign 13 were analyzed by collaborator Lauren Biddle to measure stellar rotation periods for six of the targets for which rotation periods from ground-based data could not be determined. A generalized Lomb-Scargle periodogram (Zechmeister &

K¨urster2009) of each of the lightcurves was computed. A search for 10,000 points between

0.042 and 80 days (consistent with the Nyquist sampling frequency; Press et al. 1992) was done to ensure that the peaks in the periodogram were well resolved. Distinct periodic signals of rotation for each target were identified, with no strong peaks found at periods beyond 20 days. The rotation periods were determined from the maximum peaks. Rebull et al. (2020) recently presented rotation periods, also derived from the K2 Campaign 13 data, which are in agreement with these measurements. V710 Tau A is an exception: this work measured a period of ∼4.0 days instead of the 4.3-days from Rebull et al. (2020).

The period uncertainties were estimated from the FWHM of the peak, corresponding to the frequency resolution (Ivezi´cet al. 2014). Both the analytic solution (Zechmeister &

K¨urster2009) and a Monte Carlo bootstrap algorithm were used to calculate the false- alarm probabilities (FAPs). Both methods yielded FAPs of <10−6 for all relevant periods.

Not all photometric observations determined a conclusive rotation period due to quasiperiodic variability. Irregular photometric variations in CTTSs primarily result from accretion from the circumstellar disk onto the star that varies in both time and location

(Herbst et al. 2007). Because of these limitations, rotation periods for several active

97 accretors among the CTTSs are not reported because the lightcurves only show transient periodic behavior (AA Tau, CW Tau, DK Tau, FM Tau, FZ Tau, GM Aur, HN Tau, HO

Tau, IQ Tau).

3.2.5 Stellar Radius Limits

Given the measured v sin i and rotation periods, I was able to place lower limits on the stellar radius using the relation:

v sin i P R sin i = rot . (3.1) (2π)

I calculated this value for those stars that have both v sin i and Prot estimates and propagated the errors from v sin i and Prot, where available (many Prot estimates from the literature did not include uncertainties). I used the infrared v sin i measurements to determine these values.

3.3 Discussion

3.3.1 Distributions of Stellar Parameters

Table 3.2 lists the Teff , infrared and optical v sin i,Prot, and R sin i measurements for the sample. Figure 3.4 shows a comparison between the various stellar properties. As expected,

the v sin i shows a strong anti-correlation with the stellar rotation period (Figure 3.4b).

There are no obvious correlations among the other stellar parameters. This hints at a more complicated relationship between evolutionary state and rotation than for normal main sequence stars (see Section 3.3.4). The typical Teff and R sin i is 3800 K and >1 R , respectively, as expected for low-mass pre-main sequence stars contracting down onto the main sequence.

98 Figure 3.4 Comparison of the stellar properties measured in this work. Upward arrows −1 indicate limits for fast rotators (>50 km s ). The Prot error bars are smaller than the points in the figures.

99 Table 3.1. K-S and A-D Test Results

Comparison N K-S A-D p-value p-value

single/multiple v sin i 40/29 0.056 0.054 infrared/optical v sin i 26/26 0.440 0.405 infrared/optical CTTS v sin i 19/19 0.526 0.336 infrared/optical WTTS v sin i 7/7 0.938 0.990 infrared/published v sin i 32/32 0.795 0.812 infrared/published CTTS v sin i 21/21 0.531 0.530 infrared/published WTTS v sin i 11/11 0.374 0.634 CTTSs/WTTS v sin i 43/21 0.004 0.004 Nguyen CTTSs/WTTS v sin i 54/90 0.177 0.179 combined CTTSs/WTTS v sin i 97/111 0.033 0.018

Note. — Summary of the K-S and A-D test results for various v sin i comparisons in the sample. The number of measurements and the p-values for both statistical tests are also listed.

3.3.2 Effects of Multiplicity on v sin i

Measurements of v sin i in binary stars can differ from single stars either through a measurement bias (line blending) or physical effects such as tidal-spin up (Attridge & Herbst

1992). The physical separations of the multiples in the sample span a wide range between

<10 au and several hundred au (Kraus et al. 2012). Only ∼1/3 of these systems are expected to be resolved by IGRINS. Kraus & Hillenbrand (2009) report K band contrast ratios between 0.7-0.9 for ∼1/4 of these unresolved systems. This indicates that there may be a measurement bias toward higher v sin i in the sample of multiple stars.

To determine if these broadening effects are present in the data, I compared the average v sin i of the stellar binaries (or multiples) in the sample with that of the single stars. The average v sin i of the 28 pre-main sequence stars in multiple star systems is 19.3 ± 1.8 km s−1, while the 34 single stars yield an average v sin i of 15.4 ± 1.7 km s−1, indicating that these samples are statistically different.

100 A Kolmogorov-Smirnov (K-S) and Anderson-Darling (A-D) test were used to confirm this result. The two-sample K-S test is a non-parametric method that compares two samples and determines whether they come from a single population. The k-sample A-D test is a modification of the K-S test and makes a similar comparison, however, the A-D test gives more weight to the tails of the distributions. These statistical tests determine a p-value, the probability of obtaining a result assuming that the distributions are the same. Both a K-S test and A-D test show that these two distributions are different with a marginally significant p-value of 0.056 and 0.054, respectively. The distributions are shown in Figure

3.5 and the details of the statistical tests are shown in Table 3.1.

I looked for evidence of contamination from a stellar companion for those targets that are part of a binary or multiple system and that have a higher v sin i than the sample average.

First, I looked for obvious signs of companion contamination, such as a double-peak in the

CCF. Another potential indicator of line blending from a stellar companion, particularly for equal-mass binaries, is temporal variation in the measured v sin i that could mimic

rotational broadening (Stauffer et al. 1997). Spectra of long-period binaries may display

wide lines that vary in width instead of distinct double lines. To check for this effect in

the data, I investigated the variation in the measured v sin i values of the relevant targets.

Three of the four observations of the CTTS binary, IT Tau (v sin i = 37±2 km s−1), show

a double-peak in the CCF indicating potential contamination from a stellar companion. IT

Tau is a known double star (Mathieu 1994) with an angular separation of ∼2.5 mas and

a K band contrast ratio of 0.9 (Kraus & Hillenbrand 2009). The primary and companion

should be barely resolved with this observing system, so it is possible there is contamination

from the secondary. The WTTS binary LkHα 332/G1 (v sin i = 29±2 km s−1) shows large

differences in the v sin i measurements that could be temporal variation caused by a stellar

companion. Based on separations measured by Kraus et al. (2012), I determined that this

binary is not resolved with the observing system. I measured larger than average variations

in the v sin i values for some CTTSs, regardless of multiplicity, which can be attributed to

veiling and low SNR of the spectral lines. However, these effects are not expected to impact

101 Figure 3.5 The v sin i distribution of single and multiple system pre-main sequence stars in the sample. A K-S test and A-D test confirm that these distributions are marginally statistically different. This indicates that the v sin i measurements are higher on average for stars with stellar companions, which could be a consequence of contamination and line blending in the spectra, shorter disk lifetimes, or of tidal interactions in hierarchical triple systems.

102 the v sin i measurements of the WTTSs that are noted above. There was no direct evidence for contamination in the other multiple system spectra.

The above analysis suggests that the higher v sin i values for pre-main sequence binaries compared to single stars can likely not be entirely explained by measurement bias.

Alternatively, it is possible that evolutionary factors may explain why stars in binary or multiple systems have a higher v sin i on average than those in single systems. Close binaries can be spun up by tidal interactions, leading to faster rotation rates (Attridge & Herbst

1992), and may experience truncation of the circumstellar disk, which would otherwise slow rotation (Stauffer et al. 2018). Therefore, components of close binaries generally have rotation rates that are faster than those of single stars at young ages, however wider binaries are expected to have rotation rates similar to single stars (Stauffer et al. 2018). Most of the companions in the sample have been confirmed using adaptive optics imaging (Kraus et al.

2012), and are therefore too widely separated to experience tidal spin-up. It is possible that some of these binaries are actually hierarchical triples composed of a close inner binary pair and an outer companion. Tokovinin et al. (2006) determined that the fraction of spectroscopic binaries with an additional tertiary companion is ∼60-70%, making these types of systems relatively common. Evolutionary factors involving a companion close enough to influence the primary circumstellar disk may be responsible for the higher than average v sin i of some of the targets.

3.3.3 Comparison of Optical and Infrared v sin i

Comparison to Optical Values

A direct comparison is made between v sin i values measured at optical and infrared

regimes. In addition to the difference in wavelengths, these v sin i values were calculated

from observations taken during different epochs and using different instruments. The

analysis techniques also differed; the optical measurements were made using observed stellar

templates and the LSD technique, and the infrared measurements relied on synthetic stellar

models (see Section 3.2.2 and 3.2.3 for analysis details). A direct comparison of the results

103 from two different methods at two different wavelengths can confirm the error estimates or identify astrophysical reasons for discrepancies between the two datasets. Figure 3.6 shows a comparison between the optical and infrared v sin i values and the residuals. The v sin i values above the estimated upper limit for fast rotators were excluded from the comparison.

In general, there is good agreement between these two datasets. The most significant outlier

(bottom center) is XZ Tau, which is known to be heavily veiled (White & Ghez 2001). While veiling does not broaden or narrow the FWHM of the CCF, the continuum emission caused by warm dust weakens the stellar absorption lines resulting in a CCF that lacks clear structure. This can lead to an inaccurate v sin i calculation. Published results from Nguyen et al. (2012) determined a v sin i value of ∼15 km s−1 for XZ Tau, which falls in between the infrared value of ∼18 km s−1 and the optical value of ∼10 km s−1.

A statistical comparison of the 26 infrared and optical v sin i values within the established measurement limits does not show a statistically significant difference, with a p-value of

0.440 from a K-S test and a p-value of 0.405 from an A-D test (Table 3.1). This result can place limits on potential causes of variations between the infrared and optical measurements caused by the method or the wavelength regime. One key difference is that synthetic spectra were used for the infrared analysis and observed templates for the optical. Since observed templates are non-zero rotators, it is expected that this difference in the method would result in underestimated values in the optical measurements, particularly at low values of v sin i, however, this was shown to be a negligible effect (Section 3.2.3). The observed wavelength regimes may also be responsible for some of the variation. Infrared observations could be affected by contamination of the CO lines from nonstellar line emission caused by high velocity material around the star (Johns-Krull & Valenti 2001). For the CTTSs that host disks of warm dust, the infrared v sin i measurements may be impacted by this emission. Studies have found that near-infrared spectra of CTTSs include a contribution of

CO, which affects the same CO lines at ∼2.3 µm that were used in the v sin i analysis (Carr

1989; Calvet et al. 1991). While there are various possible sources of this CO emission, one possibility is that it comes from the circumstellar disk (Scoville et al. 1983; Thompson 1985).

104 Figure 3.6 Comparison of the optical and infrared v sin i measurements and their residuals. The most significant outlier (XZ Tau, bottom center) is a heavily veiled target. A K-S test and an A-D test do not find a statistical difference between the infrared and optical v sin i measurements.

105 Spectroscopy of CO emission from young stars indicates that it often has the characteristic spectral shape of emission from a rotating disk (Carr et al. 1993; Najita et al. 1996; Chandler et al. 1995). Depending on the mass accretion and irradiation rates of the system, the effect of the CO contribution from the disk either decreases the strength of the absorption lines in the CO bands or converts them into emission (Calvet et al. 1991). Carr (1989) found that the near-infrared spectra of young stars with outflows are characterized by emission in the

CO lines, however, these bands are either absent or in absorption for most of his observed targets. Despite these findings, Casali & Eiroa (1996), Johns-Krull & Valenti (2001), and

Johns-Krull (2007) have determined that, for the vast majority of CTTSs, the CO lines in the K band originate in the stellar photosphere rather than the disk.

To test whether there is a significant disk CO contribution in the CTTS systems, I also separately compared infrared and optical v sin i values for CTTSs and WTTSs (Figure

3.6). A larger difference between the measurements at the two wavelengths for the CTTS subset would indicate significant CO emission from the disks in the CTTS systems. A

K-S test of the CTTS infrared and optical v sin i values shows that they are statistically indistinguishable, with a p-value of 0.526. Similarly, the WTTS subset is indistinguishable with a p-value of 0.938. An A-D test indicates a p-value for the CTTS sub-sample of 0.336 and 0.990 for the WTTS sub-sample. Table 3.1 summarizes the K-S and A-D test results between the infrared and optical measurements. Based on this result, there is no significant evidence for CO contamination in the CTTS systems.

Comparison to Literature Values

To confirm the result in the previous section, I also made comparisons between the infrared v sin i measurements and literature values determined from optical observations from Hartmann & Stauffer (1989) and Nguyen et al. (2012). Figure 3.7 shows a comparison to the v sin i values in Hartmann & Stauffer (1989) for an intersecting sub-sample of 21 targets, and excluding targets with rotation outside the limits. There is agreement within

∼1-3 km s−1 for ∼85% of the sub-sample. For the remaining ∼15%, these targets were

106 identified as either being heavily veiled (DL Tau), residing in a multiple system (HK Tau), or having a v sin i near the resolution limits of the two surveys (DM Tau). Many of the v sin i estimates from Hartmann & Stauffer (1989) do not have uncertainties, and a K-S test and A-D test do not show conclusive agreement between the two results with p-values of

0.135 and 0.207, respectively.

I compared the infrared v sin i measurements to those from Nguyen et al. (2012) for the

intersecting subset of 32 targets. Table 3.2 lists the Nguyen et al. (2012) v sin i values. K-S

and A-D tests indicate that the two sub-samples of v sin i measurements come from the

same parent distribution, with p-values of 0.795 and 0.812, respectively (Table 3.1). The

optical measurements from this work are mostly consistent with the Nguyen et al. (2012)

values for an intersecting subset of 22 targets: 55% of the sub-sample matches to within 1-σ,

while 91% matches to within 3-σ, however, a K-S and A-D test do not determine conclusive

agreement between the samples with p-values of 0.175 and 0.086, respectively.

To further investigate whether the infrared v sin i measurements are significantly affected

by a contribution of CO from the disk, I compared the infrared and Nguyen et al. (2012)

values separately for the 21 CTTSs and 11 WTTSs in the overlapping sample (Figure 3.8).

To confirm whether there is indeed a better match for WTTSs, I repeated the statistical

tests on both the CTTS and WTTS sub-samples. A K-S and A-D test of the CTTS infrared

and optical v sin i values shows that the samples are statistically indistinguishable with a

p-value of 0.531 and 0.530, respectively (Table 3.1). This result matches the K-S test

result for the infrared and optical CTTS v sin i comparison. For the WTTS subset, the

distributions are statistically the same with a p-value of 0.374 (K-S test) and 0.634 (A-D

test) (Table 3.1). Therefore, a comparison between the infrared v sin i values and published

optical v sin i values, shows no evidence that CO contamination from the disks significantly

affects the infrared v sin i measurements.

107 Figure 3.7 Comparison of the infrared v sin i measurements with published optical measurements from the literature. A K-S and A-D test indicate that the samples are not statistically different.

108 Figure 3.8 Comparison of the infrared v sin i measurements with published values from Nguyen et al. (2012). A K-S and A-D test indicate that the infrared and published v sin i measurements are statistically the same for the CTTS sample, but this result is less conclusive for the WTTS sample.

3.3.4 Correlation of v sin i and Evolutionary States

CTTS and WTTS Distributions

Previous studies have investigated whether there is a significant difference between the v sin i distributions of CTTSs and WTTSs (Walter et al. 1988; Hartmann & Stauffer 1989; Bouvier et al. 1993b; Edwards et al. 1993; Choi & Herbst 1996; Rhode et al. 2001; Herbst et al. 2002;

Sicilia-Aguilar et al. 2005; Nguyen et al. 2009). CTTSs are prevented from spinning up as they contract due in part to magnetic disk-braking, but WTTSs no longer have a disk (or only have a tenuous disk) and spin-up as they contract. Therefore, stellar evolution theory

109 predicts that stars with disks rotate more slowly than those without (Dahm et al. 2012), resulting in three scenarios: slow rotators with disks, slow rotators without disks that have not yet spun-up, and fast rotators without disks. For the most part, identifying fast rotators with disks is unexpected (Nguyen et al. 2009), although there can be exceptions, such as young sources that have not yet established disk-locking (Lamm et al. 2005; Cauley et al.

2012), stars that have been spun-up from a brown dwarf merger (Armitage & Bonnell 2002), or exceptions based on stellar magnetic field strength and disk accretion rates (Johns-Krull

& Gafford 2002). Many previous studies have found a statistically significant difference between the v sin i distribution of CTTSs and WTTSs, with WTTSs rotating faster than

CTTSs (Walter et al. 1988; Bouvier et al. 1993b; Edwards et al. 1993; Choi & Herbst 1996;

Herbst et al. 2002; Sicilia-Aguilar et al. 2005), yet other surveys have found that these distributions are actually statistically the same (Hartmann & Stauffer 1989; Rhode et al.

2001) or that the results are inconclusive (Nguyen et al. 2009).

A histogram comparing the CTTS and WTTS v sin i distributions of the sample is shown in the top panel of Figure 3.9. The distributions suggest that there are populations of slow- rotating stars with disks, slow-rotating stars without disks, and fast-rotating stars without disks, as expected by stellar evolution theory. There are a few exceptions in the CTTS distribution that have v sin i > 20 km s−1, which may be due to the reasons noted above.

The WTTSs in this sample are generally faster rotators than the CTTSs. A K-S test shows that the distributions are statistically different with a p-value of 0.004; an A-D test confirms this result with an identical p-value (Table 3.1). While this outcome supports the idea of v sin i evolution of the pre-main sequence stars in the sample, the sample selection may also explain this result. The sample was primarily chosen to include slow rotators (v sin i

< 20 km s−1), and additional targets were chosen from a sample predominantly comprised of WTTSs. Therefore, the target selection ensures that the fast rotators in the sample are more likely to be WTTSs rather than a randomly chosen subset of CTTSs and WTTSs.

As shown in Section 3.3.3, a comparison between the measured v sin i values and those from Nguyen et al. (2012) statistically match for a subset of overlapping targets. Therefore,

110 Figure 3.9 The v sin i distributions for CTTSs and WTTSs in the sample (top panel), a sample chosen from Nguyen et al. (2012) (middle panel), and the combination of both samples (bottom panel). All v sin i values above the upper limit of 50 km s−1 are plotted at this limit. A K-S and A-D test show that these distributions are statistically different in this work. The sample from the literature does not show conclusive evidence of this difference, while the combined sample indicates that the v sin i distributions for CTTSs and WTTSs are statistically distinguishable. Therefore, this difference in the v sin i distributions is likely caused by a selection bias.

111 to further explore this result, I increased the sample size by including 54 additional CTTSs and 90 WTTSs from Nguyen et al. (2012). Only the targets with v sin i < 50 km s−1 were included since this is the upper limit on the v sin i that can be accurately measured in this work. The middle panel of Figure 3.9 shows the Nguyen et al. (2012) v sin i distributions for CTTSs and WTTSs. The chosen subset from Nguyen et al. (2012) does not conclusively show that the CTTS and WTTS v sin i distributions are different, with a K-S test and A-D test indicating a p-value of 0.177 and 0.179, respectively (Table 3.1). However, combining the subset of additional v sin i measurements with the values from this work (bottom panel of

Figure 3.9) results in two distributions that are statistically distinguishable, with a p-value of

0.033 (K-S test) and 0.018 (A-D test) (Table 3.1). This result reinforces the conclusion that the statistically significant difference in the CTTS and WTTS v sin i distributions detected in this work is likely due to selection bias and is not seen in the Nguyen et al. (2012) results, however, the combined dataset also shows a statistically significant difference between the v sin i distributions of the two populations.

Evolution on the H-R Diagram

The v sin i of a pre-main sequence star is expected to change as the star evolves towards the main sequence, first by decreasing due to interactions with the circumstellar disk, then by increasing after the disk disperses and the star continues to contract.

To investigate the correlation between the v sin i and stellar evolution, I plotted the sample on a Hertzsprung-Russell diagram (H-R diagram) and a near-infrared color- magnitude diagram (CMD), which is sensitive to the presence of a disk, in Figure 3.10.

I collected H and K magnitudes and parallaxes, from which distances and absolute magnitudes could be calculated, from the Simbad database1 for ∼70% of the sample and combined them with the Teff measurements. Overplotted on both diagrams is a subset of Pleiades targets (∼100 Myr old; Meynet et al. 1993), for reference. These stars are primarily

1http://simbad.u-strasbg.fr

112 on the zero-age main sequence. The Pleiades H and K magnitudes were measured by Bouy et al. (2015) and the Gaia DR2 distance estimates were determined by Abramson (2018).

The H and K magnitudes of stars in young systems may be affected by extinction caused by the molecular cloud or circumstellar disks. In the CMD, the spread of the CTTSs to the right of the zero-age main sequence may be caused by the presence of circumstellar disks in the CTTS systems (Figure 3.10). These disks can cause reddening and strong infrared excess emission, as indicated by the brighter K band magnitudes (and subsequent higher

H-K color index).

Figure 3.10 shows no obvious relation between H-R diagram position, v sin i and the presence of a disk, indicating a complex interplay between stellar rotation and the evolution of pre-main sequence stars towards the zero-age main sequence. While there are no obvious trends, there is tentative evidence that the objects closest to the zero-age main sequence are the fastest and slowest rotators in the sample (dark squares and light circles, respectively).

The apparent evolution of low-mass stars to the zero-age main sequence does not proceed at the same rate for all pre-main sequence stars, and WTTSs are not always older than their CTTS counterparts (Walter et al. 1988). However, Walter et al. (1988) found that the timescale on which disks dissipate appears to vary widely on a timescale comparable to the time a star spends on the Hayashi track, and is dependent on local initial conditions. The fast and slow rotators closest to the zero-age main sequence seemingly support this finding.

Shorter disk lifetimes also lead to naturally faster rotating stars since disks magnetically lock to the star and prevent spin-up as the star contracts. The position of the fast-rotating

WTTSs near the zero-age main sequence suggests that these WTTSs preferentially began evolving towards the zero-age main sequence as they lost their disks. As noted in Section

2.1, there is a selection bias that favors WTTSs for the sample of fast rotating stars (v sin i

> 20 km s−1).

The slow rotators close to the zero-age main sequence are CTTSs, which supports the

finding that slow rotating stars in young clusters are those with long disk lifetimes and low initial v sin i (Bouvier 1997), and could be evidence of evolved CTTSs with significantly

113 Figure 3.10 A Hertzsprung-Russell diagram (top) and a color-magnitude diagram (bottom) with the ∼50 pre-main sequence stars in the sample that have measurements of H and K magnitudes, parallaxes, and Teff . This sample includes CTTSs and WTTSs with the v sin i indicated by color. Stars from the Pleiades open cluster are overplotted to demonstrate the evolutionary stage of the pre-main sequence star sample relative to the zero age main sequence. The CTTSs display reddening from extinction and strong infrared excess emission from their circumstellar disks, as indicated by the brighter K band magnitudes in the color-magnitude diagram. Typical error bars are shown in the upper left (the absolute K-magnitude error is smaller than the symbol sizes).

114 decreased v sin i caused by magnetic braking from long-lived disks. The grouping of the slowest rotating CTTSs near the zero-age main sequence may further provide evidence that these systems are evolved; they may have disks that are beginning to disperse and therefore are not as significantly reddened by extinction as the other CTTSs in the sample (although any extinction from a molecular cloud would still be present).

It is important to note that inferring evolutionary states from the position of pre- main sequence stars on the H-R diagram or CMD is limited. Magnitudes are affected by photometric variability caused by accretion (Messina et al. 2017), stellar luminosity and Teff are affected by starspots (Somers & Pinsonneault 2015; Flores et al. 2019), and Li abundance as an age indicator, as well as other astrophysical measurements, are complicated by radius inflation (Somers & Pinsonneault 2014). These phenomena all have an effect on the placement of pre-main sequence stars on evolutionary tracks.

3.4 Conclusions

This chapter presents stellar properties for 70 low-mass pre-main sequence stars in the

Taurus-Auriga region, including projected rotational velocities, effective temperatures, rotation periods, and limits on radii. About 70% of the sample show evidence of still having an accreting circumstellar disk and are classified as CTTSs, and just under half the sample are known multiples. The Teff was calculated using infrared spectra and Teff -LDR relations (L´opez-Valdivia et al. 2019), v sin i was calculated using cross-correlation analysis of infrared and optical spectra, Prot was determined from ground-based and space-based photometry and published values, and R sin i was estimated from a relation between v sin i and Prot. The main scientific conclusions are as follows:

• Infrared v sin i measurements of pre-main sequence stars calculated using the cross

correlation technique agree with those measured by an independent MCMC technique,

with a typical scatter of ∼1.8 km s−1. This defines a typical error floor for the v sin i

of these stars from infrared spectra.

115 • A comparison of the v sin i distributions of stars with and without companions shows

a marginally significant difference, with binaries/multiples typically having higher

measured v sin i. This can partly be explained by line broadening from companion

spectral lines. Tidal interactions in hierarchical triples may also affect some of these

systems.

• A comparison of optical and infrared v sin i values shows no statistical difference

regardless of whether the star has a disk or not. This indicates that CO contamination

from the disk does not significantly impact v sin i measurements above the typical error

floor of the measurements (∼1.8 km s−1).

• There is no clear correlation between v sin i, presence of a disk, and H-R diagram

position, which indicates a complex interplay between stellar rotation and evolution

of pre-main sequence stars. There is tentative evidence that the objects closest to

the zero-age main sequence are fast-rotating WTTSs and slow-rotating CTTSs. The

former supports the finding that pre-main sequences stars that evolve quickly towards

the main sequence have short disk lifetimes and become fast rotators without the

presence of a disk to prevent spin-up. The latter may be evidence of evolved CTTSs

that have become slow rotators due to magnetic disk braking of long-lived disks. This

result indicates that the evolution of pre-main sequence stars is both complex and

varied.

The results presented here illustrate the importance of high-resolution infrared spectroscopy to characterize pre-main sequence stars. These stellar properties were determined as part of the infrared RV survey to find and characterize young hot Jupiters and other substellar companions.

116 Table 3.2 Stellar Properties of Pre-Main Sequence Stars

a b c d e f g Target NIR/Nopt Multiplicity Teff Infrared v sin i Optical v sin i Published v sin i Prot R sin i C/W −1 −1 −1 (K) (km s ) (km s ) (km s ) (days) (R ) AA Tau 16/91 ··· 3800 13.1± 1.8 12.3 ± 0.9 12.8 ± 1.1 ······ C1 BP Tau 7/421 ··· 3900 9.1 ± 2.0 8.5 ± 0.7 13.1 ± 1.6 7.0 ± 0.11 1.3 ± 0.3 C1 CI Tau 51/231 ··· 4200 12.0 ± 1.8 10.1 ± 0.7 ··· 6.6 ± 0.11 1.6 ± 0.2 C1 CoKu LkHα 332 G2 4/0 binary 3900 23.1 ± 1.9 ············ W1 CoKu Tau 3 4/0 binary 3800 8.7 ± 1.8 ············ C1 CW Tau 6/21 ······ 29.3 ± 3.4 28.1 ± 0.4 ········· C1 CZ Tau 5/0 binary 3400 ········· 5.5 ± 0.11 ··· C1 DE Tau 6/0 ··· 3700 10.6 ± 1.8 ··· 9.7 ± 0.3 7.64 1.6 ± 0.3 C1 DG Tau 3/101 ··· 4200 ··· 24.3 ± 1.0 24.7 ± 0.7 ······ C1 DH Tau 22/81 ··· 3700 8.5 ± 1.8 7.7 ± 0.8 10.9 ± 0.6 5.2 ± 0.11 0.9 ± 0.2 C1 DI Tau 5/101 binary 3900 13.8 ± 1.9 11.7 ± 0.7 12.5 ± 0.6 7.7 ± 0.11 2.1 ± 0.3 W1 DK Tau 32/191 binary 3900 15.1 ± 1.8 13.6 ± 0.8 17.5 ± 1.5 ······ C1 DL Tau 7/71 ··· 4300 9.3 ± 1.9 7.5 ± 0.8 19 ± 4 ······ C1 117 DM Tau 5/0 ··· 3600 6.6 ± 1.8 ··· 4.0 ± 0.7 15.24 2.0 ± 0.5 C1 DN Tau 3/431 ··· 3900 10.8 ± 1.8 9.8 ± 0.7 12.3 ± 0.6 6.3 ± 0.11 1.3 ± 0.2 C1 DR Tau 5/101 ··· 4000 ··· 7.0 ± 4.0 ··· 15.7 ± 0.11 ··· C1 DS Tau 19/161 ··· 4000 12.9 ± 1.8 10.7 ± 0.8 ········· C1 FM Tau 6/0 ··· 3500 9.7 ± 1.8 ············ C1 FZ Tau 5/0 ··· 3700 6.7 ± 1.9 ············ C1 GG Tau A 4/0 binary 4000 11.4 ± 1.8 ··· 11.5 ± 0.7 ······ C1 GI Tau 5/21 ··· 3900 11.3 ± 1.9 8.5 ± 0.8 12.7 ± 1.9 7.1 ± 0.11 1.6 ± 0.3 C1 GK Tau 5/151 ··· 4200 19.8 ± 1.8 20.1 ± 0.8 ··· 4.6 ± 0.11 1.8 ± 0.2 C1 GM Aur 9/121 ··· 4300 13.7 ± 1.7 13.5 ± 0.8 14.8 ± 0.9 ······ C1 GM Tau 5/0 ··· <3100 15.9 ± 1.8 ······ 2.68 ± 0.042 0.8 ± 0.1 C2 GN Tau 5/0 binary 3600 12.7 ± 1.8 ············ C1 GV Tau 4/0 binary 4500 22.6 ± 1.9 ············ C1 Haro 6-37 4/11 binary 4200 13.8 ± 1.8 12.4 ± 2.8 12.1 ± 1.2 ······ C1 HK Tau 5/0 binary 3800 22.1 ± 1.9 ······ 3.3 ± 0.12 1.4 ± 0.1 C1 HN Tau 4/0 binary ·················· C1 HO Tau 4/0 ··· 3600 21.1 ± 1.8 ············ C1 IP Tau 6/11 ··· 3800 10.3 ± 1.8 10.8 ± 1.1 10.6 ± 0.4 ······ C1 IQ Tau 7/161 ··· 3800 13.5 ± 1.8 13.4 ± 0.8 14.4 ± 0.3 ······ C1 continued on the next page Table 3.2, continued.

a b c d e f g Target NIR/Nopt Multiplicity Teff Infrared v sin i Optical v sin i Published v sin i Prot R sin i C/W −1 −1 −1 (K) (km s ) (km s ) (km s ) (days) (R ) IRAS F04113+2758 4/0 binary 3500 4.1 ± 2.1 ············ C3 IRAS F04192+2647 3/0 ··· 3800 12.4 ± 2.0 ············ C4 IRAS F04248+2612 4/0 binary 3400 24.1 ± 1.9 ············ C3 IRAS F04325+2402 4/0 binary ·················· C3 IRAS F04370+2559 4/0 ··· 3800 14.0 ± 1.9 ············ C1 IRAS F04385+2550 3/0 ··· 4000 23.7 ± 1.9 ············ C1 IT Tau 4/0 binary 4700 37.4 ± 2.0 ······ 2.74 ± 0.042 2.0 ± 0.1 C1 IW Tau 7/321 binary 3800 9.2 ± 1.8 8.6 ± 0.7 8.7 ± 0.8 5.5 ± 0.11 1.0 ± 0.2 W1 L1551 IRS 5 4/0 ··· 3400 31.2 ± 1.9 ············ C5 L1551-55 5/0 ··· 4200 7.3 ± 1.8 ··· 7.7 ± 0.7 ······ W1 L1642-2 3/0 binary 3700 14.4 ± 1.8 ············ C6 LkCa 4 7/91 ··· 3700 29.6 ± 1.8 27.9 ± 1.3 10.6 ± 0.4 3.45 2.0 ± 0.1 W1 1 6 1 118 LkCa 5 6/2 binary 3700 36.8 ± 1.9 39.5 ± 0.8 ··· 1.4 1.02 ± 0.05 W LkCa 15 10/101 ··· 4900 14.6 ± 1.8 13.6 ± 0.8 13.9 ± 1.2 5.77 1.6 ± 0.2 C1 LkHa 332/G1 5/0 binary 3700 28.7 ± 2.3 ············ W1 NTTS 040142+2150 NE 3/0 binary 3400 16.0 ± 1.9 ············ W6 NTTS 040142+2150 SW 3/0 binary 3400 36.3 ± 1.9 ············ W6 NTTS 040234+2143 3/0 ··· 3500 6.1 ± 1.9 ············ W6 NTTS 041559+1716 3/31 ······ >50 >50 74 ± 4 0.7 ± 0.11 >0.7 W6 NTTS 043230+1746 3/0 ··· 3600 5.0 ± 1.8 ············ W6 RW Aur 4/32 binary 4100 16.0 ± 2.2 19.4 ± 5.1 ··· 5.08 1.6 ± 0.2 C1 RX J0425.3+2618 3/0 quadruple 4000 11.9 ± 1.8 ············ W7 T Tau 5/0 triple 4500 20.0 ± 2.1 ··· 23.0 ± 1.2 ······ C6 UX Tau B 5/0 binary 3600 11.0 ± 1.8 ············ W1 V410 Tau 4/31 binary ··· >50 >50 83 ± 4 1.95 >1.8 W1 V710 Tau A 3/0 binary 3700 21.4 ± 1.9 ··· 21.5 ± 0.4 4.05 ± 0.092 1.7 ± 0.2 C6 V710 Tau B 3/0 binary 3700 17.4 ± 1.8 ··· 18.31 ± 0.19 ······ W6 V807 Tau 4/0 triple 3400 13.0 ± 2.6 ··· 13.6 ± 0.7 4.4 ± 0.12 1.1 ± 0.2 C8 V819 Tau 5/0 ··· 4000 10.0 ± 1.8 ··· 9.1 ± 0.6 5.55 1.1 ± 0.2 C1 V830 Tau 54/121 ··· 4500 32.5 ± 1.8 32.6 ± 1.0 32.0 ± 1.5 2.7 ± 0.11 1.8 ± 0.1 W1 V836 Tau 6/131 ··· 3800 12.1 ± 1.8 12.5 ± 0.9 13.4 ± 1.1 6.8 ± 0.11 1.6 ± 0.2 C1 V928 Tau 4/0 binary 4000 34.1 ± 1.8 ··· 31.6 ± 0.7 2.3 ± 0.11 1.5 ± 0.1 W1 continued on the next page Table 3.2, continued.

a b c d e f g Target NIR/Nopt Multiplicity Teff Infrared v sin i Optical v sin i Published v sin i Prot R sin i C/W −1 −1 −1 (K) (km s ) (km s ) (km s ) (days) (R ) V1075 Tau 54/211 ··· 4200 32.3 ± 1.8 33.1 ± 0.8 32.1 ± 1.4 2.4 ± 0.11 1.5 ± 0.1 W1 V1095 Tau 4/0 ··· 3600 34.2 ± 1.8 ··· 30.9 ± 1.1 2.5 ± 0.11 1.7 ± 0.1 W1 V1096 Tau 4/51 ··· 4000 ··· 17.7 ± 0.9 ··· 6.5 ± 0.11 ··· W6 V1115 Tau 3/151 ··· 4200 23.0 ± 1.8 23.4 ± 0.8 16.65 ± 0.04 3.3 ± 0.11 1.5 ± 0.1 W1 XZ Tau 6/51 binary 3500 18.3 ± 1.8 10 ± 3 15.0 ± 1.2 ······ C1 ZZ Tau 4/0 binary 3400 20.6 ± 2.0 ······ 4.2 ± 0.12 1.7 ± 0.2 C1 119 aIn the cases where the target name represents a wide binary, only the primary component was observed and characterized, unless the secondary is also specified. bObservations are from the (1) 107-inch telescope or the (2) 82-inch telescope, both at McDonald Observatory. c The relations used to determine Teff are limited to values between 3100-4100. All values outside this range are extrapolated. d The Teff uncertainties are 200 K. eThe published v sin i come from Nguyen et al. (2012). f The stellar rotation periods were obtained from (1) Lowell Observatory photometry, (2) K2 data, (3) Bouvier et al. (1995), (4) Artemenko et al. (2012), (5) Grankin et al. (2008), (6) Norton et al. (2007), (7) Donati et al. (2019), (8) Percy et al. (2006). gEach target is classified as C (classical T Tauri star) or W (weak-line T Tauri star). These values came from the literature: (1) Kraus et al. (2012) (and references therein), (2) Monin et al. (2010), (3) Andrews and Williams (2005), (4) Akeson et al. (2019), (5) Adams et al. (1987), (6) Herbig & Bell (1988), (7) Chavarr´ıa-Ket al. (2000), (8) Schaefer et al. (2012). References

Abramson, G. 2018, Research Notes of the AAS, 2, 150

Allard, F. & Hauschildt, P. H. 1995, in The Bottom of the Main Sequence - and Beyond,

ed. C. G. Tinney, 32

Armitage, P. J. & Bonnell, I. A. 2002, MNRAS, 330, L11

Attridge, J. M. & Herbst, W. 1992, ApJ, 398, L61

Bastian, N., Kamann, S., Cabrera-Ziri, I., Georgy, C., Ekstr¨om,S., Charbonnel, C., de

Juan Ovelar, M., & Usher, C. 2018, MNRAS, 480, 3739

Benbakoura, M., R´eville,V., Brun, A. S., Le Poncin-Lafitte, C., & Mathis, S. 2019, A&A,

621, A124

Blackman, E. G. & Thomas, J. H. 2015, MNRAS, 446, L51

Bolmont, E., Raymond, S. N., Leconte, J., & Matt, S. P. 2012, A&A, 544, A124

Bouvier, J. 1990, AJ, 99, 946

—. 1997, Mem. Soc. Astron. Italiana, 68, 881

Bouvier, J. 2013, in EAS Publications Series, Vol. 62, EAS Publications Series, ed.

P. Hennebelle & C. Charbonnel, 143–168

Bouvier, J., Bertout, C., Benz, W., & Mayor, M. 1986, A&A, 165, 110

120 Bouvier, J., Cabrit, S., Fernandez, M., Martin, E. L., & Matthews, J. M. 1993a, A&A, 272,

176

—. 1993b, A&AS, 101, 485

Bouvier, J. & C´ebron,D. 2015, MNRAS, 453, 3720

Bouy, H., Bertin, E., Sarro, L. M., Barrado, D., Moraux, E., Bouvier, J., Cuillandre,

J. C., Berihuete, A., Olivares, J., & Beletsky, Y. 2015, VizieR Online Data Catalog,

J/A+A/577/A148

Browning, M. K., Basri, G., Marcy, G. W., West, A. A., & Zhang, J. 2010, AJ, 139, 504

Calvet, N., Patino, A., Magris, G. C., & D’Alessio, P. 1991, ApJ, 380, 617

Camenzind, M. 1990, Reviews in Modern Astronomy, 3, 234

Carr, J. S. 1989, ApJ, 345, 522

Carr, J. S., Tokunaga, A. T., Najita, J., Shu, F. H., & Glassgold, A. E. 1993, ApJ, 411, L37

Casali, M. M. & Eiroa, C. 1996, A&A, 306, 427

Cauley, P. W., Johns-Krull, C. M., Hamilton, C. M., & Lockhart, K. 2012, ApJ, 756, 68

Chandler, C. J., Carlstrom, J. E., & Scoville, N. Z. 1995, ApJ, 446, 793

Chen, W. & Johns-Krull, C. M. 2013, ApJ, 776, 113

Choi, P. I. & Herbst, W. 1996, AJ, 111, 283

Clarke, C. J. & Bouvier, J. 2000, MNRAS, 319, 457

Cohen, O., Drake, J. J., Kashyap, V. L., Sokolov, I. V., & Gombosi, T. I. 2010, ApJ, 723,

L64

Cummings, J. D. & Kalirai, J. S. 2018, AJ, 156, 165

121 Dahm, S. E., Slesnick, C. L., & White, R. J. 2012, ApJ, 745, 56

Davison, C. L., White, R. J., Henry, T. J., Riedel, A. R., Jao, W. C., Bailey, J. I., I., Quinn,

S. N., Cantrell, J. R., Subasavage, J. P., & Winters, J. G. 2015, AJ, 149, 106

Deen, C. P. 2013, AJ, 146, 51

Donati, J. F., Semel, M., Carter, B. D., Rees, D. E., & Collier Cameron, A. 1997, MNRAS,

291, 658

D’Orazi, V., Biazzo, K., & Randich, S. 2011, A&A, 526, A103

Edwards, S., Strom, S. E., Hartigan, P., Strom, K. M., Hillenbrand, L. A., Herbst, W.,

Attridge, J., Merrill, K. M., Probst, R., & Gatley, I. 1993, AJ, 106, 372

Flores, C., Connelley, M. S., Reipurth, B., & Boogert, A. 2019, ApJ, 882, 75

Gaia Collaboration. 2018, VizieR Online Data Catalog, I/345

G´alvez-Ortiz, M. C., Clarke, J. R. A., Pinfield, D. J., Jenkins, J. S., Folkes, S. L., Garc´ıa

P´erez,A. E., Day-Jones, A. C., Burningham, B., Jones, H. R. A., Barnes, J. R., &

Pokorny, R. S. 2010, MNRAS, 409, 552

Gameiro, J. F. & Lago, M. T. V. T. 1993, MNRAS, 265, 359

Goorvitch, D. 1994, ApJS, 95, 535

Gray, D. F. 2008, The Observation and Analysis of Stellar Photospheres

Harris, A. W., Young, J. W., Bowell, E., Martin, L. J., Millis, R. L., Poutanen, M., Scaltriti,

F., Zappala, V., Schober, H. J., Debehogne, H., & Zeigler, K. W. 1989, , 77, 171

Hartmann, L., Hewett, R., Stahler, S., & Mathieu, R. D. 1986, ApJ, 309, 275

Hartmann, L. & Stauffer, J. R. 1989, AJ, 97, 873

Hartmann, L. W. & Noyes, R. W. 1987, ARA&A, 25, 271

122 Herbst, W., Bailer-Jones, C. A. L., Mundt, R., Meisenheimer, K., & Wackermann, R. 2002,

A&A, 396, 513

Herbst, W., Eisl¨offel,J., Mundt, R., & Scholz, A. 2007, in Protostars and Planets V, ed.

B. Reipurth, D. Jewitt, & K. Keil, 297

Houk, N. & Swift, C. 1999, Michigan Spectral Survey, 5, 0

Ivezi´c, Z.,ˇ Connelly, A. J., Vand erPlas, J. T., & Gray, A. 2014, Statistics, Data Mining,

and Machine Learning in Astronomy

Johns-Krull, C. M. 2007, ApJ, 664, 975

Johns-Krull, C. M. & Gafford, A. D. 2002, ApJ, 573, 685

Johns-Krull, C. M., McLane, J. N., Prato, L., Crockett, C. J., Jaffe, D. T., Hartigan, P. M.,

Beichman, C. A., Mahmud, N. I., Chen, W., Skiff, B. A., Cauley, P. W., Jones, J. A., &

Mace, G. N. 2016, ApJ, 826, 206

Johns-Krull, C. M. & Valenti, J. A. 2001, ApJ, 561, 1060

Johnstone, C. P. 2017a, A&A, 598, A24

Johnstone, C. P. 2017b, in IAU Symposium, Vol. 328, Living Around Active Stars, ed.

D. Nandy, A. Valio, & P. Petit, 168–179

Johnstone, C. P., G¨udel,M., St¨okl,A., Lammer, H., Tu, L., Kislyakova, K. G., L¨uftinger,

T., Odert, P., Erkaev, N. V., & Dorfi, E. A. 2015, ApJ, 815, L12

Kastner, J. H., Huenemoerder, D. P., Schulz, N., Zuckerman, B., Weintraub, D. A., Gaidos,

E. J., & Negoro, H. 1997, in American Astronomical Society Meeting Abstracts, Vol. 191,

92.06

Kesseli, A. Y., Muirhead, P. S., Mann, A. W., & Mace, G. 2018, AJ, 155, 225

123 Kislyakova, K., Lammer, H., Erkaev, N., Holmstr¨om,M., Odert, P., Khodachenko, M.,

Kulikov, Y., & G¨udel,M. 2013, in Protostars and Planets VI Posters

Koenigl, A. 1991, ApJ, 370, L39

Kraus, A. L. & Hillenbrand, L. A. 2009, ApJ, 704, 531

Kraus, A. L. & Hillenbrand, L. A. 2009, The Astrophysical Journal, 704, 531

Kraus, A. L., Ireland, M. J., Hillenbrand, L. A., & Martinache, F. 2012, ApJ, 745, 19

Lamm, M. H., Mundt, R., Bailer-Jones, C. A. L., & Herbst, W. 2005, A&A, 430, 1005

L´opez-Valdivia, R., Mace, G. N., Sokal, K. R., Hussaini, M., Kidder, B. T., Mann, A. W.,

Gosnell, N. M., Oh, H., Kesseli, A. Y., Muirhead, P. S., Johns-Krull, C. M., & Jaffe,

D. T. 2019, ApJ, 879, 105

Lyubchik, Y., Jones, H. R. A., Pavlenko, Y. V., Pinfield, D. J., & Covey, K. R. 2012,

MNRAS, 422, 2195

Mathieu, R. D. 1994, ARA&A, 32, 465

—. 2003, arXiv e-prints, astro

Matt, S. & Pudritz, R. E. 2005a, ApJ, 632, L135

—. 2005b, MNRAS, 356, 167

Messina, S., Parihar, P., & Distefano, E. 2017, MNRAS, 468, 931

Meynet, G., Mermilliod, J. C., & Maeder, A. 1993, A&AS, 98, 477

Najita, J., Carr, J. S., Glassgold, A. E., Shu, F. H., & Tokunaga, A. T. 1996, ApJ, 462, 919

Nguyen, D. C., Brandeker, A., van Kerkwijk, M. H., & Jayawardhana, R. 2012, The

Astrophysical Journal, 745, 119

124 Nguyen, D. C., Jayawardhana, R., van Kerkwijk, M. H., Brandeker, A., Scholz, A. e., &

Damjanov, I. 2009, ApJ, 695, 1648

Nordhagen, S., Herbst, W., Rhode, K. L., & Williams, E. C. 2006, AJ, 132, 1555

Pecaut, M. J. & Mamajek, E. E. 2013, ApJS, 208, 9

Piskunov, N. 1999, in Astrophysics and Space Science Library, Vol. 243, Polarization, ed.

K. N. Nagendra & J. O. Stenflo, 515–525

Piskunov, N. E., Kupka, F., Ryabchikova, T. A., Weiss, W. W., & Jeffery, C. S. 1995,

A&AS, 112, 525

Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery, B. P. 1992, Numerical recipes

in C. The art of scientific computing

Rebull, L. M., Stauffer, J. R., Cody, A. M., Hillenbrand , L. A., Bouvier, J., Roggero, N.,

& David, T. J. 2020, AJ, 159, 273

Reiners, A., Joshi, N., & Goldman, B. 2012, AJ, 143, 93

Rhode, K. L., Herbst, W., & Mathieu, R. D. 2001, AJ, 122, 3258

Ryabchikova, T. & Pakhomov, Y. 2015, Baltic Astronomy, 24, 453

Scoville, N., Kleinmann, S. G., Hall, D. N. B., & Ridgway, S. T. 1983, ApJ, 275, 201

Shu, F., Najita, J., Ostriker, E., Wilkin, F., Ruden, S., & Lizano, S. 1994, ApJ, 429, 781

Sicilia-Aguilar, A., Hartmann, L. W., Szentgyorgyi, A. H., Fabricant, D. G., F˝ur´esz, G.,

Roll, J., Conroy, M. A., Calvet, N., Tokarz, S., & Hern´andez,J. 2005, AJ, 129, 363

Sills, A., Pinsonneault, M. H., & Terndrup, D. M. 2000, ApJ, 534, 335

Smith, C. 2015, PhD Thesis, Georgia State University

Soderblom, D. R., Pendleton, J., & Pallavicini, R. 1989, AJ, 97, 539

125 Sokal, K. R., Deen, C. P., Mace, G. N., Lee, J.-J., Oh, H., Kim, H., Kidder, B. T., & Jaffe,

D. T. 2018, ApJ, 853, 120

Somers, G. & Pinsonneault, M. H. 2014, ApJ, 790, 72

—. 2015, ApJ, 807, 174

Soto, M. G. & Jenkins, J. S. 2018, A&A, 615, A76

Spruit, H. C. 2018, arXiv e-prints, arXiv:1810.06106

Stauffer, J., Rebull, L. M., Cody, A. M., Hillenbrand, L. A., Pinsonneault, M., Barrado,

D., Bouvier, J., & David, T. 2018, AJ, 156, 275

Stauffer, J. R., Balachandran, S. C., Krishnamurthi, A., Pinsonneault, M., Terndrup, D. M.,

& Stern, R. A. 1997, ApJ, 475, 604

Sun, W., de Grijs, R., Deng, L., & Albrow, M. D. 2019, ApJ, 876, 113

Thompson, R. I. 1985, ApJ, 299, L41

Tinker, J., Pinsonneault, M., & Terndrup, D. 2002, ApJ, 564, 877

Tokovinin, A., Thomas, S., Sterzik, M., & Udry, S. 2006, A&A, 450, 681

Tout, C. A. & Pringle, J. E. 1992, MNRAS, 256, 269

Valenti, J. A. & Fischer, D. A. 2005, VizieR Online Data Catalog, J/ApJS/159/141

Vidotto, A. A. 2016, in 19th Cambridge Workshop on Cool Stars, Stellar Systems, and the

Sun (CS19), Cambridge Workshop on Cool Stars, Stellar Systems, and the Sun, 147

Vogel, S. N. & Kuhi, L. V. 1981, ApJ, 245, 960

Walter, F. M., Brown, A., Mathieu, R. D., Myers, P. C., & Vrba, F. J. 1988, AJ, 96, 297

White, R. J. & Ghez, A. M. 2001, ApJ, 556, 265

126 Wolff, S. C., Strom, S. E., & Hillenbrand, L. A. 2004, ApJ, 601, 979

Xiao, H. Y., Covey, K. R., Rebull, L., Charbonneau, D., Mandushev, G., O’Donovan, F.,

Slesnick, C., & Lloyd, J. P. 2012, ApJS, 202, 7

Zechmeister, M. & K¨urster,M. 2009, A&A, 496, 577

127 Chapter 4

A Search for Substellar Companions around

Pre-Main Sequence Stars

In addition to pre-main sequence star characterization, the infrared spectroscopy provides radial velocities for the substellar companion search. Periodic variation in the

RVs may be indicative of a companion orbital period, although other possible sources must also be explored. This chapter introduces the RV methodology (Section 4.1), including a description of the technique used for measuring pre-main sequence star RVs (Section 4.1.1), the RV program algorithms and various modifications that were made (Section 4.1.2), and the survey precision (Section 4.1.3). Also included is a discussion of the RV analysis used to search for periodic signals (Section 4.1.4), sources of spurious signals (Section 4.1.5), and injection/recovery simulations used to determine survey sensitivity and upper limits on occurrence rates (Section 4.1.6). The results of the RV companion search for individual systems are presented in Section 4.2.1. Section 4.2.2 discusses the survey sensitivity, upper limits on companion occurrence rates, infrared RV scatter in comparison to the optical, and possible correlations between RV scatter and system properties.

128 4.1 Methodology

4.1.1 Forward Modeling Technique

Radial velocities were measured from the infrared spectroscopy using a forward modeling technique (Blake et al. 2010; Crockett et al. 2011; Johns-Krull et al. 2016). The observed spectra contain both stellar photospheric absorption lines and telluric absorption lines from the Earth’s atmosphere. This method fits a combined stellar model and telluric spectrum to the observed spectrum and measures the shift of the stellar lines relative to the telluric lines. Figure 4.1 illustrates this technique for a sub-region of the IGRINS K band data.

The K band region, centered around 2.3 µm, has numerous narrow and deep CO absorption lines originating in the stellar atmosphere of low-mass stars, which are suitable for accurate

RV analysis. This region also contains sharp telluric absorption lines, primarily caused by

CH4 in the Earth’s atmosphere.

An alternative method to the forward modeling technique is cross correlation analysis, which was used to determine the v sin i (Section 3.2.2). While the width of the cross correlation function indicates the v sin i, the shift of the function peak indicates the RV.

The forward modeling technique offers two advantages over the cross correlation method:

1) it measures relative RVs rather than absolute RVs, so the science and calibration spectral

lines are equally affected by the uncertainty in the wavelength solution (1 pixel or 2 km

s−1), and effects such as instrument flexure, and therefore do not need to be corrected, and

2) this method results in higher precision than cross correlation analysis (see Mann et al.

(2016a,b) for additional details on IGRINS RVs using cross correlation analysis). A full

discussion of the survey precision is presented in Section 4.1.3.

To measure a RV from the shift in the spectral lines, a wavelength reference must

be established in the spectrum. Many instruments employ a gas absorption cell for this

calibration, whereas the spectral lines of the gas are overlaid on the spectra and serve as

the wavelength reference. This is a challenging approach in the infrared regime, however,

because no known gas creates the dense set of absorption lines needed to serve as a suitable

129 Figure 4.1 Illustration of the forward modeling technique used to determine RVs. The figure shows a sub-region of the K band spectra used in this analysis. A model spectrum comprised of a synthetic stellar spectrum (a) and an observed telluric template (b) is created. The model is then fit to the observed spectrum (red), shown in (c). Residuals are shown in (d). The shift of the stellar lines relative to the telluric lines indicates the measured RV.

130 reference within a wide wavelength range in the infrared. This is the case with Thorium-

Argon lamps, which are successfully used at visible wavelengths, but are ineffective in the K band region. A combination of gas cells or the addition of an emission lamp have also been used for infrared RV calibration (see Mahadevan & Ge (2009), Reiners et al.

(2010), and Cale et al. (2019) for additional details on the infrared calibration problem and proposed solutions). An ammonia gas cell has also been employed as a wavelength reference for infrared spectra, however, limitations in the ammonia spectral line coverage and overlapping telluric features make it less than ideal for wavelength calibration (Bean et al. 2010). In the place of a gas cell, telluric lines can serve as a simultaneously-observed wavelength reference. Absorption from the Earth’s atmosphere is prevalent in the near- infrared (Blake et al. 2010; Figueira et al. 2010a,b; Bean et al. 2010) and the atmosphere is estimated to be stable to about 10 m s−1 (Figueira et al. 2010a). Therefore, telluric lines provide a significant number of OH, H2O, and CH4 lines for relatively stable wavelength calibration. This calibration technique has also been used for other RV surveys (Seifahrt &

K¨aufl2008; Blake et al. 2010; Bean et al. 2010; Muirhead et al. 2011; Crockett et al. 2011,

2012; Johns-Krull et al. 2016).

4.1.2 RV Measurements

CSHELL Pipeline

To measure RVs using the forward modeling technique, I used a series of IDL programs

developed for spectra from the infrared spectrograph, CSHELL on the NASA Infrared

Telescope Facility. An accompanying program that modifies IGRINS data to be compatible

with the CSHELL code was also used. Variations of the CSHELL programs were used for

RV analysis in Mahmud et al. (2011); Crockett et al. (2011, 2012); Johns-Krull et al. (2016).

The CSHELL program implements the following steps:

• Set a wavelength range for analysis of a single order in the K band,

avoiding the edges of the spectrum where dispersion is significant. The

131 program analyzes a spectral region within a single order in the K band (the example

spectra from Section 2.2.2 shown in Figure 2.8 presents an example of the full spectral

order used in this analysis). The CSHELL analysis uses nearly the full spectral order

(between 2.2973 µm and 2.3214 µm).

• Fit a polynomial to the continuum and normalize the spectrum. The

spectrum is normalized by dividing the spectrum flux and uncertainties by a

polynomial continuum fit.

• Combine a synthetic stellar spectrum and telluric template to create a

single model to fit to the spectrum. The RV fitting model is constructed

by multiplying a synthetic stellar spectrum by a telluric template. Section 3.2.2

discussed the stellar models used in the v sin i and RV analysis, which were generated

by collaborator, Chris Johns–Krull. Essentially, a model stellar atmosphere was

generated by interpolating over a grid of NextGen stellar atmosphere models (Allard

& Hauschildt 1995) tailored to an assumed Teff (4000 K, which corresponds to the

average Teff of the RV targets), log g (3.5, see Section 3.2.2), and metallicity (assumed to be solar) for the sample. The SYNTHMAG spectrum synthesis program (Piskunov

1999) was used to compute the stellar template spectrum using line data from from the

Vienna Atomic Line Database (Piskunov et al. 1995; Ryabchikova & Pakhomov 2015).

The telluric template came from telluric absorption spectra extracted from K band

observations of the Sun made by the National Solar Observatory (NSO) (Livingston

& Wallace 1991). Figure 4.1 shows the stellar model (a) and telluric template (b).

• Fit the model to the spectrum and optimize a series of fitting parameters,

including the stellar RV. The CSHELL program fits the combined model spectrum

to the observed spectrum by optimizing a set of user-specified parameters. The input

parameters are: 1) velocity shifts (stellar RV and telluric RV), 2) scaling factors for

both RV inputs, 3) v sin i, 4) instrumental profile full width at half maximum (IP

FWHM), which is a measure of spectral line broadening caused by the instrument, 5)

132 four continuum fitting coefficients (to normalize the spectrum) based on a third order polynomial fit, and 6) three wavelength solution coefficients (to account for dispersion) based on a second order polynomial fit.

The stellar RV is the primary parameter of interest and I estimated the starting input values based on those reported by spectroscopic surveys of the Taurus-Auriga region

(Torres et al. 2006; Nguyen et al. 2012; Gaia Collaboration 2018). The telluric RV describes the velocity of the Earth’s atmosphere, which can show small variations due to effects like atmospheric turbulence. Since the telluric lines are used as the wavelength reference, this parameter remains fixed at zero in order to allow the program to optimize the wavelength solution. Both velocities have scaling factors associated with them, which are also free parameters. The stellar RV and telluric

RV scaling factors describe the depth of the spectral features. The optimized velocity shifts and scaling factors are applied to the stellar and telluric templates, and the stellar spectrum is then interpolated onto the telluric template wavelength scale.

In order to match the observed spectrum, the model spectrum has to be artificially broadened. The rotation of the star broadens the stellar spectral lines (characterized by the v sin i) and the instrument causes broadening to both the telluric and stellar lines (characterized by the IP FWHM). The v sin i parameter remains fixed, as this significantly increases the efficiency of the fitting procedure. The applied v sin i values correspond to those measured in Chapter 3. The v sin i is used to artificially apply rotational broadening to the stellar synthetic spectrum to match the data, prior to multiplying the stellar model with the telluric template to produce the total model spectrum. The limb darkening coefficient, which determines the rotationally broadened profiles, is held fixed at 0.3, based on a linear limb-darkening law (Gray

2008).

133 The IP FWHM parameter, a measure of the instrumental resolution in pixels, was

estimated by measuring the FWHM of select telluric lines in the A0 standard star

observations. This value was typically ∼3 pixels, or ∼5 pixels during the final

observing season when the K band resolution was lower (see Section 2.2.2 for details).

After the combined model is generated, the model spectral lines are convolved by the

optimized IP FWHM parameter.

Finally, the model is re-binned to the parameterized wavelength scale characterizing

the dispersion. This wavelength solution was optimized by fitting a second order

polynomial to the wavelength scale and optimizing the coefficients. Likewise, the third

order polynomial fit for the continuum normalization (noted above) is optimized and

applied to the model spectrum. The optimization of these parameters allow the model

to be fit to the observed spectrum so the stellar RV shift can be measured.

The CSHELL program optimizes the parameters using the Levenberg-Marquardt

method (or damped least-squares method), which solves non-linear least squares

problems by optimizing the reduced chi-squared (Bevington & Robinson 1992).

• Determine a random uncertainty for the RV using Monte Carlo

simulations. For each observation, 100 simulated spectra are generated by adding

noise to the best fit model. The noise is scaled by weights based on spectrum

flux uncertainties. The model is re-fit to the 100 simulated observations and the

standard deviation of the resulting output is taken as the random uncertainty for that

observation. RV random uncertainties are typically on the order of tens of m s−1,

with some observations resulting in 100-200 m s−1.

• Correct the output RV for the motion of the barycenter (the center of

mass of the Solar System). Once the parameters are optimized, the final stellar

RV is determined based on the model fit to the observed spectrum and the barycentric

correction. For this correction, the barycentric velocity was calculated for a particular

observing site, time, and target. This velocity accounts for the geocentric motion of

134 the observatory, the heliocentric motion of the Earth, and the barycentric motion

of the Sun. The barycentric velocity is added to the final stellar RV to convert the

measurement to the reference frame of the barycenter. The program outputs the

optimized parameters, including the barycentric-corrected stellar RV, the RV random

uncertainty, the SNR of the observations, the reduced chi-squared, and the mean and

standard deviation of the best fit residuals. The standard deviation of the best fit

residuals is typically 2-3%.

Prior to receiving the CSHELL programs, I developed an independent code as part of this dissertation, also written in IDL. The programs both functioned in similar ways with some key differences: the region chosen for analysis and the method used for parameter optimization.

Both programs use the same single order in the K band, but the independent program I developed uses a region within this order that was ∼20% smaller than that analyzed by the

CSHELL program. The smaller region avoided the significant telluric absorption features between 2.316-2.320 µm (Figure 2.8) that can disrupt the continuum fitting. The continuum

fit is implemented in the same way in both programs, however, it was possible to obtain a more accurate continuum fit using the smaller region. Including a larger region requires compromises on one end of the spectrum or the other because of the telluric features noted above, which artificially cause a dip in the continuum in that region.

My program implemented the IDL function, AMOEBA, an optimization routine that uses the downhill simplex method (or amoeba method). This method takes a starting input for independent variables and uses geometric relationships to find a minimum, terminating the routine when it reaches a threshold value. It is a derivative-free optimization that relies only on function evaluations (Nelder & Mead 1965).

Both the CSHELL program and the independent program are similarly constructed and perform essentially the same functions. My program provided a better fit to the observed spectrum, as measured by chi-squared, however it results in larger RV variations between observations and larger random uncertainties. This is primarily due to the smaller spectral

135 region used, where there are fewer stellar and telluric lines. The CSHELL program better

fit the larger spectral regions than the independent program, and the increased number of spectral lines proved to be very important for measuring accurate RVs. A comparison of results from the CSHELL program and the independent program is shown in Figure 4.2.

The RVs of a standard star, GJ 281, which are expected to remain constant over time, were measured. Testing a limited subset of observations revealed that the small region tested in the independent program provided better precision than the original CSHELL program, with precisions of ∼50 and ∼100 m s−1, respectively. However, expanding the spectral region in the independent program resulted in a precision >200 m s−1, and modifications

to the CSHELL program (discussed in Section 4.1.2) increased the precision to ∼70 m s−1. A full discussion of the RV precision of this survey is discussed in Section 4.1.3.

Additionally, the capabilities of the CSHELL program had been tested on a known hot

Jupiter around Gl 86, a quiescent main sequence star (Queloz et al. 2000). As noted in

Crockett et al. (2011, 2012) the planet was confirmed using this program, thus proving the accuracy of RVs measured with the CSHELL program. A similar IGRINS dataset was not immediately available to test the independent program. Ultimately, I determined that the benefit of including additional lines was more important than a lower chi-square fit to a smaller spectral region given the improved accuracy. This choice was confirmed by the Gl

86b recovery results. The RV analysis and results discussed in this dissertation use the RVs measured from the CSHELL programs, including the modifications outlined below.

Pipeline Modifications

I made several modifications to the CSHELL programs prior to completing the analysis.

The changes are outlined below:

• Wavelength range: The single order K band spectrum used in this analysis was

trimmed to avoid the increased distortion and decreased signal to noise at the

spectrum edges. Approximately 0.004 µm were removed on either edge, which

removed the most regions most affected by distortion while still maintaining the

136 Figure 4.2 A comparison of the CSHELL programs (squares) and the independent RV programs (circles) using a limited RV stable star dataset. The original CSHELL program (blue) achieved a precision of ∼100 m s−1 and the small region of the independent program (green) resulted in a precision of ∼50 m s−1. However, a larger region was necessary to increase the RV accuracy, and the independent program did not provide good fits with larger spectral regions (red). The modified CSHELL program (orange) was chosen to complete the analysis presented in this dissertation.

137 largest number of spectral lines possible to maintain accuracy. Figure 4.3 shows the

full spectral order used in this analysis and the final trimmed region.

• Spectrum flux uncertainty: These values represent the uncertainty in the

measured spectrum flux based on noise in the observations, and serve as weights

for the fitting routine. The prior uncertainty estimate was made by subtracting the

continuum fit from the spectrum flux, which was multiplied by the instrument gain,

then added to the square of the read noise. The spectrum flux uncertainty can instead

be estimated from the SNR. I modified the program to determine a flux uncertainty

by dividing the spectrum flux by the SNR. The modification resulted in decreased

flux uncertainties overall, which were typically lower by a factor of 3-4.

• Final stellar RV determination: The program initially defined the final RV based

on the Monte Carlo random uncertainty simulation. After fitting 100 simulated

spectra, 100 RVs from the simulations were measured and averaged to determine

the final RV. Because of the random noise generated in the simulations, this process

resulted in a slightly different fit, and therefore different RV output, each time the

program was run. I measured the RV separately from the Monte Carlo simulation by

fitting the observed spectrum rather than taking the average RV from the simulated

spectra fits.

• Random uncertainty determination: Initially, the Monte Carlo simulation

relied on the observed spectrum to estimate the random uncertainties. During the

simulations, scaled noise was added to the observed spectrum for each of the 100

simulations. I modified this simulation to use the best fit model instead of the

observed spectrum. By adding noise to the best fit spectrum and re-fitting it, the

random uncertainty based on the noise estimate is better quantified than by re-fitting

the observed spectrum with additional noise.

• Spectrum continuum fit: Because the program was initially designed for CSHELL

spectra, the continuum fitting procedure had to be modified for the IGRINS spectra.

138 The program analyzes a spectral order centered near ∼2.3 µm for both datasets,

however, the IGRINS orders are shifted relative to the CSHELL orders. The

wavelength range included in the corresponding CSHELL order is largely within the

order overlap in the IGRINS data. The IGRINS data, therefore, includes heavy

telluric absorption features that are not part of the CSHELL spectral order (Figure

2.8). These features make continuum fitting more challenging. To account for this, I

generalized the program to do an initial second order continuum fit, followed by a third

order fit with coefficients that are optimized in the fitting program. The improvement

made to the continuum fit is shown in Figure 4.4.

• Minor debugging: I reviewed the IDL programs called by the main program and

corrected any inconsistencies in parameter format between the variables that were

passed and the formats required by the called programs.

• RVs for higher mass stars: I modified the program to analyze a spectral region

in the H band to measure RVs of higher mass stars (with spectral type A, F, or

G). There were limitations in finding a suitable spectral region that had both strong

stellar and telluric features to measure RVs of stars of higher temperatures. Without

these features, it is difficult to accurately measure the RV. This result highlights the

challenge of measuring stellar RVs of higher mass stars using the forward modeling

technique. Numerous, narrow absorption lines from both the star and Earth’s

atmosphere are needed to accurately measure the RV. These features are much more

prevalent in K band observations of cool stars. I therefore limited my sample to stars

of spectral type K and M.

Several of these modifications to the program were made in an effort to improve the survey precision, which is discussed in the next section.

139 Figure 4.3 An illustration of the modified spectral region used in this analysis. The edges of the spectral order were trimmed by ∼0.004 µm to avoid distortion. The full order spectrum is shown in black and the region used in the analysis is shown in red.

140 Figure 4.4 An illustration of the improved continuum fit from the modified CSHELL program. The IGRINS K band order used in this analysis includes heavy telluric features between ∼2.316-2.320 µm which complicates the continuum fit. The initial continuum- correction was made using a second order polynomial fit (black spectrum). In the modified CSHELL program, the program implements an initial second order continuum fit, followed by a third order fit. This resulted in a distinct improvement in the continuum correction (green). A horizontal blue dotted line is included in the figure to provide context for the final continuum fit and normalization.

141 4.1.3 RV Precision

IGRINS Survey Precision

The typical RV precision of the survey was determined through observations of an RV standard star, GJ 281. This is a low-mass star (spectral type = g, K mag = 5.9, Teff = 3776 K; Casagrande et al. 2008) that has a stable RV, with a velocity dispersion of 6.6 m s−1

(Endl et al. 2003). Any scatter in the RVs can primarily be attributed to the observational

system and method, and serves as an estimate of the reproducibility of the measurements.

The GJ 281 RVs were measured in the same way as the targets, as outlined in Section 4.1.2.

The standard star RVs are known and serve as an additional test confirming the accuracy of

the modified CSHELL code. The standard deviation of the GJ 281 RVs provides an estimate

of a method uncertainty, which also serves as an indication of the survey sensitivity and

detection limits. The method error added in quadrature with the random error gives the

total uncertainty on a RV measurement. The 64 GJ 281 observations (JD, RV, and random

uncertainty) are listed in Table A.1 in Appendix A. Figure 4.5 shows the relative RVs

for the 3 seasons of IGRINS/LDT data measured with the modified CSHELL RV program

(top). The final season, which had lower resolution (see Section 2.2.2 for details), shows

more RV variation than the previous seasons.

The short-term RV variability of GJ 281 is very low, resulting in a precision of ∼30-40

m s−1 over days or weeks. However, over a longer timescale, such as the survey timescale

of months or years, the precision decreases, due in part to limitations of long-term stability

for the observing system. A measure of the initial standard deviation of the GJ 281 RVs

was 132 m s−1 over all observing seasons prior to the follow-up season. I completed several

tests to determine if I could improve the precision to better than 100 m s−1.

The successful tests included trimming the edges of the spectrum to avoid the regions with the most distortion and improving the continuum fit (see Section 4.1.2 for details).

Removing ∼0.004 µm on either edge of the order increased the precision to ∼75 m s−1.

Improving the continuum fit to account for the heavy telluric absorption features also

142 Figure 4.5 RVs of the standard star, GJ 281 collected during the three IGRINS/LDT observing seasons and normalized by the mean from each season. The top panel shows the RVs vs. time measured using the modified CSHELL RV program. The final season presents more scatter because of the lower instrument resolution. The bottom panel shows the RVs from the first two seasons that were measured prior to improving the RV program and survey precision. It is apparent that there is less RV scatter in the first two seasons of the modified measurements when the IGRINS spectral resolution was higher.

143 modestly increased the precision to ∼68 m s−1. These tests are illustrated in Figures

4.3 and 4.4 and resulted in an improved precision from ∼130 m s−1 to ∼70 m s−1 for all

observing seasons prior to the follow-up season. An additional test improved the fit, but

did not change the precision: at the survey resolution, several of the telluric lines in the

analysis region are saturated, so I also imposed a threshold to mask saturated lines from the

fitting routine that reach below a normalized flux of 0.8. While this did not change the RV

results, it did improve the fit, as indicated by the reduced chi-squared output from the RV

measurement program. The RVs measured with the improved CSHELL program are shown

in Figure 4.5 (top), and the RVs from the first two seasons that were measured prior to the

improvements are also shown (bottom). The RV scatter is reduced in the measurements

made with the modified RV program.

I completed several additional tests to potentially improve the precision, however, these

did not result in increased sensitivity overall. To account for possible errors introduced by

small inconsistencies in the wavelength solution, I imposed a single wavelength scale on each

spectrum. I also allowed the telluric lines to vary during the optimization routine instead

of remaining fixed. In an effort to improve the continuum fit even further, I converted

the spectra into log space where the curve of the spectrum would be less pronounced, and

therefore potentially easier to fit. I also experimented with trimming the spectrum further,

but removing additional spectral lines to avoid regions with lower SNR and increased

distortion, did not significantly improve the output and resulted in lower accuracy overall.

To complete the final precision improvement test, I worked with an undergraduate

student, Cassandra Bodin. This test explored whether including additional spectral regions

in the RV analysis would increase the survey sensitivity by increasing the number of spectral

lines used in the calculation. Possible spectral regions were identified from an infrared solar

spectrum atlas that included both spectral lines from a stellar photosphere with starspots

and telluric lines (Livingston & Wallace 1991). No suitable regions for RV analysis were

identified in the H band, however, three additional K band regions were flagged as having

strong absorption line features conducive to RV analysis (2.324-2.356 µm, 2.354-2.387 µm,

144 and 2.385-2.418 µm). The associated spectral orders have absorption lines that are fewer in number and less narrow and deep than the original analysis region, and the stellar lines are also CO dominated. Collaborator Chris Johns–Krull provided synthetic stellar spectra for these additional orders and I created telluric templates from the Livingston & Wallace

(1991) NSO dataset. Cassandra Bodin tested the CSHELL RV analysis program on these new spectral regions, removing data that were poorly fit (with a chi-squared cutoff of 100).

The precision for each additional region was determined to be ∼200 m s−1. Combining the

orders did not improve the precision, as there was a systematic offset between the IGRINS

spectral orders. Therefore, while a given precision could be achieved within a single order,

adding orders to the analysis would not increase the precision without first characterizing

that offset. The analysis was instead completed with the single original spectral order

centered around 2.3 µm.

The final single-order precision estimate based on the standard deviation of the GJ 281

RVs was determined to be 68 m s−1 for all data collected prior to the follow-up observing

season. Because of lower resolution in the K band, the final observing season has an

estimated precision of 115 m s−1. A combination of the full dataset gave a precision of 95

m s−1. Since this precision estimate is based on observations of a quiet main sequence star,

it does not account for the intrinsic RV variability inherent in young stars. This precision

served as an estimate of the IGRINS method uncertainty, and was added in quadrature

with the random uncertainty of each RV estimate. The random uncertainties of the survey

pre-main sequence stars range from several tens of m s−1 to between 100-200 m s−1.

A precision of ∼70-120 m s−1 is adequate for a survey searching for short-period, high-

mass, young substellar companions. Young stars are expected to exhibit an intrinsic infrared

RV variability >100 m s−1 (Mahmud et al. 2011), so higher precision measurements are not

required to detect close-in high-mass planetary or brown dwarf companions around these

stars. To put this precision into context of the survey sensitivity, Figure 4.6 illustrates the

estimated detection limits for various precisions with previously detected planets from the

145 NASA Exoplanet Archive1. The diagram shows companion mass vs. distance from the host star. The hot and warm Jupiter regimes, assuming a mass of at least 1 MJ, are outlined in red and blue, respectively. Various precisions are indicated by the diagonal lines. For the general observing seasons this survey achieved a precision of 68 m s−1, while for the follow- up observing season the precision was 115 m s−1. All companions above a given diagonal line are theoretically detectable by the survey at that precision. This survey, therefore, is expected to be sensitive to close-in brown dwarfs and planetary objects with high mass.

Detection limits based on injection/recovery tests are presented in Section 4.1.6, which give an estimate of the survey sensitivity for each target.

Comparison to the Literature

A precision threshold at the 50-100 m s−1 level is typically necessary to detect and characterize exoplanets. Other surveys have demonstrated precisions better than 100 m s−1 with a similar approach. Blake et al. (2010) and Bailey et al. (2012) used telluric absorption features as a wavelength reference in the same wavelength regime and determined a limiting precision of ∼50 m s−1 for M dwarf observations using NIRSPEC on Keck II.

Bailey et al. (2012) indicated a precision of ∼80-170 m s−1 for stars younger than 10

Myr, with observations of fast rotators resulting in decreased sensitivity. Similar CSHELL surveys estimate long-term precisions of ∼70-90 m s−1 (Crockett et al. 2012; Davison et al.

2015). The high-resolution spectrograph, iSHELL, has achieved a precision of 5 m s−1 over a one-year timescale for M and K dwarfs (Cale et al. 2019). Mann et al. (2016a) confirmed a short-term stability of IGRINS data at the 50 m s−1 level, but estimated a 150-160 m s−1 precision for young stars using the cross correlation method of RV determination. The initial precision limit of ∼130 m s−1 is comparable to this value, and also to the precision estimated for the optical spectroscopic data included in this work (see Section 2.3 and

3.2.3). For the optical RV survey, Huerta et al. (2008) and Mahmud et al. (2011) estimated a precision of ∼120-140 m s−1 using the Robert G. Tull Coud´eSpectrometer at the 2.7-m

1https://exoplanetarchive.ipac.caltech.edu

146 Figure 4.6 Estimated detection limits for various precisions. The hot and warm Jupiter regimes, assuming a mass of at least 1 MJ, are outlined in red and blue, respectively. Various precisions are indicated by the diagonal lines. This survey achieved a precision of 68 m s−1, and 115 m s−1 for the follow-up season. All companions above a given diagonal line are theoretically detectable by the survey at that precision. This survey, therefore, is expected to be sensitive to close-in brown dwarfs and planetary objects with high mass.

147 Harlan J. Smith Telescope at McDonald Observatory by measuring the overall RMS scatter of RV standard stars.

The final IGRINS precision estimate of ∼70 m s−1 (and >100 m s−1 for the follow-up

season) is in line with other estimates from surveys using a similar technique, as discussed

above. There is also an intrinsic limit to the survey sensitivity based on the analysis methods

used and the SNR of the observations. This limit is set by such factors as the theoretical

photon noise, the stability of the instrument, and variability introduced during the analysis

process. Astrophysical sources of RV noise also contribute to the overall sensitivity of a

survey. While it is difficult to quantify each noise contribution individually, several studies

have examined the variability expected from using telluric lines as a wavelength reference.

Figueira et al. (2010a) evaluated this technique and determined that telluric lines are

generally stable down to 10 m s−1 over 6 years. More specifically, Deming et al. (1987) have investigated the stability of CH4 telluric lines in the 2.3 µm region and determined that they are stable down to 20 m s−1, with the effect of winds being of the most concern.

Blake et al. (2010), however, estimate that telluric features limit RV precision to 60 m s−1.

It is expected that the most significant limiting factor of this RV survey will be the effects of stellar activity.

4.1.4 Periodogram Analysis

The process of determining RV variability caused by gravitational interactions with a companion begins with a search for periodic signals in the RV dataset. This search was implemented using the Lomb-Scargle periodogram, which is a commonly-used algorithm for detecting and characterizing periodicity in unevenly-sampled time series data (Lomb

1976; Scargle 1982). Essentially, this technique fits a sinusoidal model to the data at each frequency to create a Fourier-like power spectrum, where higher power reflects a better fit to the RV data (see VanderPlas (2018) for a detailed discussion of this method).

The periodogram analysis was done with a Lomb-Scargle module from the Astropy v4.0.1 package (Astropy Collaboration et al. 2013, 2018).

148 Prior to conducting the periodogram analysis, the mean RV was subtracted from each measurement to give the RV variation about zero rather than absolute RV values. Since

IGRINS was a visiting instrument that was re-mounted on the LDT for six-month periods over three years, there is no long-term stability. Therefore, the mean RV from each season was subtracted from the seasonal data separately. Additionally, a linear correction was also applied to the RVs to ensure the RV baseline was consistent. The lack of long-term stability does not affect a search for short-period companions (with orbital periods .100 days), but would be a limiting factor for a survey aimed at finding more distant companions.

To test the feasibility of recovering companions with close-in orbital periods in the dataset,

I simulated a power spectrum computed with the Lomb-Scargle technique (Figure 4.7).

This periodogram was generated using simulated RV data drawn from a sine curve with a

frequency corresponding to a 10-day orbital period and an amplitude of 500 m s−1. The

injected signal was combined with the time stamps from the 64 observations of the RV

standard star, GJ 281 (Figure 4.5). An uncertainty of 100 m s−1 was assumed based on

the precision estimate of the IGRINS data from the combined seasons. Noise was randomly

generated and scaled by the uncertainty, then added to the sine curve. The periodogram

analysis recovers the peak signal at the frequency set by the user to within 0.001 days.

The power spectrum for the RV dataset was computed with several user-defined

parameters. The periodogram samples frequencies corresponding to periods between 2-10

days and 10-100 days to search for signals indicating either hot or warm Jupiter candidate

detections. The 2 day lower limit is set by the Nyquist sampling limit. Observations were

generally separated by 1 day or longer, so the frequencies associated with periods less than 2

days would likely not be fully recovered in the power spectrum. The default frequency grid

is set to sample ∼5 grid points across each significant periodogram peak. This value was

increased to 10 samples per peak to more finely sample the frequency grid. RV measurement

errors are accounted for in the analysis, whereas the data points are individually weighted

by their uncertainties. This modification does not change the statistical properties of the

resulting periodogram and allows it to be constructed even in the presence of correlated

149 Figure 4.7 An example of a Lomb-Scargle periodogram created using temporal sampling from the 64 observations of the RV standard, GJ 281 and an injected sine curve corresponding to a substellar companion with a 10-day orbital period. The power spectrum successfully recovers the input period for the simulated RV data, as indicated by the maximum peak. FAP thresholds at 0.1% and 1% and are indicated by the dashed and dotted lines, respectively.

150 noise. A standard normalization by the residuals of the fit to a constant reference model is applied to the periodogram.

A false alarm probability (FAP) was calculated to determine the significance of the dominant peak detected in the power spectrum. The FAP describes the probability that a randomized dataset would measure a peak of a certain power (or higher), assuming that the data consist of only white noise, with no periodic component. The FAP was estimated using a bootstrap algorithm (Kuerster et al. 1997). This technique randomly reorders the

RVs but retains the same observed times, then re-computes the periodogram. For a FAP of

1%, this simulation is repeated 1000 times (and is repeated ∼1/FAP times for other FAP values). The FAP is then defined as the ratio of resampled periodograms with a peak at any frequency that is stronger than that of the initial detected signal. The FAP threshold was set to 1% for this analysis, corresponding to a 1% false positive rate, which describes the probability that a peak of the detected amplitude was produced by white noise. A threshold of 0.1% is also commonly used and considered to be a significant detection, however, given the inherent RV scatter in young star RV data, I chose the larger threshold of 1%, which is indicative of a statistical significance between 2-3σ. Figure 4.7 shows examples of the 0.1% and 1% FAP thresholds for the simulated dataset.

Once a significant periodic signal is identified in the RVs (as set by the FAP threshold), the data are phase-folded to the corresponding frequency (or period) and a sine wave is

fit to the phased data (Boisse et al. 2011). An example of this process for the simulated

RV data is shown in Figure 4.8. In this case, the amplitude of the simulated phased RVs matches the user input to within ∼4 m s−1. The recovered peak signal and amplitude of the simulated data demonstrate the reliability of the RV analysis.

Peak signals in the periodogram analysis do not always indicate a planet detection (see

Section 4.1.5 for more details), so a dominant signal can be removed from the dataset by subtracting the sine curve fit from the RVs. Additional periodogram analysis and phase- folding with sinusoidal fitting can then be implemented on the residuals to detect and characterize additional signals.

151 Figure 4.8 An example of simulated RVs phased to the frequency corresponding to the dominant peak in the power spectrum using the simulated RV data and RV standard star time stamps. The best-fit sinusoid is overplotted in orange and corresponds to a RV semi- amplitude of 500 m s−1.

152 4.1.5 Sources of Spurious RV Signals

Periodogram analysis can identify significant peaks in a power spectrum, however, further investigation is needed to determine whether these peaks correspond to real periodic signals or are simply artifacts of sampling. Harmonics are sometimes present at integer multiples of the true detected period and maintain a consistent power relative to each other. If the signal is non-sinusoidal, additional peaks may indicate higher frequency components of a signal. Harmonics can be caused by rotational modulation or a planet on an eccentric orbit. Similarly, observing aliases, which are spurious signals caused by gaps in observations typically at the 1-day observation interval, may also be present in the data and can be confused with a planet orbital frequency. Significant aliasing effects can be avoided by adequately observing at the full phase coverage. However, when present these aliases can cause significant peaks in a power spectrum. Noise can add coherently to an alias, or incoherently to a periodic signal, which causes the alias to appear larger. Therefore, additional tests must be applied to data to determine if peak signals are caused by aliasing rather than a companion detection (see Dawson & Fabrycky (2010) for further discussion of aliasing effects and mitigation techniques).

The spectral window function, which is a Fourier transform of a timeseries with all y- values set to 1, can be used to identify aliases. An example of a window function is shown in Figure 4.9. The GJ 281 power spectrum with the injected signal at a frequency of 0.1 days−1 is shown in the top panel, and the spectral window function corresponding to the GJ

281 timeseries is shown in the bottom panel. This window function is fairly typical of the

RV observations for this survey, as the standard star was observed with similar frequency as the follow-up targets. Peaks in the window function can combine with true frequencies causing aliases in the power spectrum, however, identifying aliases with window functions is not always straightforward, as aliases do not always appear close to the peaks in the window function.

Another possible test for aliasing is to randomize the RVs while maintaining the time stamps and taking the power spectrum. This removes the peak signal to reveal possible

153 Figure 4.9 An example of periodogram analysis of GJ 281 with an injected signal at a frequency of 0.1 days−1 (top) and the corresponding spectral window function. Window functions are used to identify significant peaks caused by aliasing from gaps in observations. 154 aliases caused by the temporal sampling. An illustration of this technique is shown in

Figure 4.10, again using GJ 281 with the injected companion signal as an example. The scrambled RVs do not show the significant peak detected in the initial periodogram analysis, which supports the conclusion that this peak is not caused by aliasing effects. There are some caveats to this approach, however. Aliases are not caused solely by the spacing of observations, but are produced by the convolution of the window function with other signals

(such as periodic signals or noise). By randomizing the RVs the peak signal is removed, which can, in some cases, remove the alias as well.

Even after excluding sampling artifacts, interpreting peak signals in a periodogram can be complex. Astrophysical sources other than planets can result in observed periodic behavior. Stellar jitter can be caused by magnetic activity, such as starspots and plages

(Saar & Donahue 1997; Desort et al. 2007; Reiners et al. 2010; Haywood 2016; Rajpaul et al. 2016). Flares can also cause short-term variability and convection creates an inhomogenous surface that can induce variations in the RVs (see Haywood (2016) for a detailed discussion). Stellar jitter is often a significant source of RV modulation, particularly in optical observations of pre-main sequence stars where stellar activity can cause RV variations much larger than most observed planetary signals (Hillenbrand et al. 2015).

Most of these activity-induced signals show periodicity on a timescale related to the stellar rotation period, and both lightcurves and RV curves are modulated by the rotation of the star. This is problematic for companion orbital periods that are close to the rotation period of the star, as it is difficult to determine the source of detected RV variability. Other astrophysical sources in pre-main sequence star systems also can create spurious periodic signals, such as changes in the disk accretion rate or extinction along the line-of-sight (Xiao et al. 2012).

All of these factors affect the detection sensitivity of a young star survey. Conducting a survey in infrared light instead of optical can reduce the effects of stellar activity, particularly starspots (see Section 2.2 for additional discussion). Combining photometry with RV monitoring is another approach to mitigating stellar activity. Several studies

155 Figure 4.10 An illustration of a technique to test for signals caused by aliasing. The periodogram (blue) shows the injected data using the GJ 281 time stamps. The time stamps are preserved and the RVs randomized to test for signals that are produced by temporal sampling (orange). If no significant peak is found in the randomized RV periodogram, the original peak signal may be astrophysical.

156 have investigated the correlation between RV and photometric variability, both to better understand the physics behind the phenomena causing the variability, and to determine how combined analysis can enhance RV survey precision (Aigrain et al. 2012; Bastien et al.

2014; Oshagh et al. 2017; Hojjatpanah et al. 2020; Kosiarek & Crossfield 2020). This survey primarily uses photometry to determine the stellar rotation period (see Section 3.2.4 for details) to identify periodicity associated with stellar magnetic activity in the power spectrum analysis. While it is possible a planet will have an orbital period matching the stellar rotation period, particularly if tidally-locked (Marcy et al. 1997), signals at the frequency of the stellar rotation are generally attributed to stellar activity and can be removed by fitting a sinusoid to the data and subtracting it from the RVs. However, noise from stellar jitter is correlated and not always strictly periodic, and therefore can be difficult to mitigate.

Several studies have applied a Gaussian process technique to RV data to disentangle stellar activity signals from planetary signals (Haywood et al. 2014; Grunblatt et al. 2015;

Rajpaul et al. 2015; Donati et al. 2016; Yu et al. 2017; Damasso et al. 2018). A Gaussian process is a non-parametric, statistical method used for modeling correlated noise by constructing comprehensive probabilistic models. This technique is able to represent stellar rotation signals that are stochastic, due in part to evolving active regions (Kosiarek &

Crossfield 2020). An example of a Gaussian process fit to simulated RV data from Rajpaul et al. (2015) is shown in Figure 4.11. For the RV dataset, I initially tested an open source

Python package, RadVel (Fulton et al. 2018), which implements a Gaussian process fit to

RV data. However, because the rotational periods of pre-main sequence stars are typically similar to the orbital periods of hot Jupiters and close-in brown dwarfs, there is a risk of overfitting the data with a Gaussian process fit, thereby removing a substellar companion signal along with the stellar activity signal. Removing the stellar rotation period from the

RV data with a standard sinusoidal fit did not introduce any significant harmonics of the subtracted sine wave in the residual dataset, indicating that this fit adequately removes the activity-induced periodicity for the purposes of this survey.

157 Figure 4.11 Example of a Gaussian process fit of simulated RVs from Rajpaul et al. (2015). A Gaussian process is a statistical method used for modeling correlated noise, such as that caused by stellar jitter.

4.1.6 Detection Limits from Injection and Recovery Tests

While periodogram analysis can enable detections of periodic signals, and additional tests can be used to investigate the source of these signals as possibly being planetary in nature, there are limitations to what any survey can detect based on limits determined by measurement uncertainties and observing cadence. Quantifying non-detections can provide insights into these young systems. By identifying the range of companion masses and orbital periods within the detection limits, it can be inferred which companions are not present in a given system, and which may be present, but are not detectable with this survey. These non- detections can provide upper limits on companion occurrence rates, and can be quantified using injection and recovery simulations to measure completeness as a function of minimum companion mass (Mp sin i) and semi-major axis. Injection and recovery tests simulate a number of planet signals in the dataset, then measures the probability of recovering the signals within the specified mass-period parameter space (Blake et al. 2010; Howard &

Fulton 2016).

Detection limits for the ten RV targets were determined using RVSearch (Howard &

Fulton 2016). RVSearch conducts an automated iterative search for planet signals in RV data, then computes planet detection limits for each star as a function of Mp sin i and semi-

158 major axis. The untargeted search served as a check on the periodogram analysis discussed in Section 4.1.4. The search uses a two-dimensional Keplerian Lomb-Scargle periodogram

(O’Toole et al. 2009), which is more sensitive to eccentric planets than the traditional

Lomb-Scargle periodogram. It also incorporates measurement errors and zero-point offsets directly into the periodogram. Because of IGRINS’ status as a visiting instrument that was re-mounted to the telescope each season, the data were separated prior to running the

RVSearch program so a distinct instrumental offset and jitter term could be defined for each season. RVSearch applies a FAP threshold to indicate potential companion detections.

The threshold is set by an empirical FAP method, which differs from the bootstrap strategy described in Section 4.1.4. The empirical FAP is derived from a linear fit of a histogram of the periodogram peaks higher than the median power value. This provides an estimate of the number of peaks above a given threshold, which is multiplied by the number of independent test periods to determine an approximate probability that a peak of a given value will be detected within the periodogram. The advantage of this technique is that correlated noise is characterized in the distribution of periodogram peak heights, which is not the case when randomizing RVs using the bootstrap method. I set the FAP threshold for detection to 1% so the RVSearch planet search had consistent criteria as the standard periodogram analysis. If a signal in the periodogram is detected above the FAP threshold, it is defined as a planet detection and the search for additional potential planet signals continues. The search automatically stops if no significant signals are detected.

Once the planet search has concluded, RVSearch injects simulated planetary RV signals into the data for each star, while preserving the observation time stamps, measured RVs, and uncertainties based on user-defined parameters. These parameters include the stellar mass, the number of simulations, the range of orbital periods and RV semi-amplitudes to sample, the eccentricity, and the FAP. The automated search algorithm is then run on the modified dataset to attempt to recover the injected signals, while also simultaneously fitting for known significant signals in the data revealed in the initial iterative search. The injected planets are on circular orbits and uniformly distributed in log space within the parameters

159 set for the RV semi-amplitude and orbital period. An injected planet signal is considered recovered if the highest peak in the periodogram is above the FAP detection threshold and within 25% of the injected period, and if the phase of the recovered orbit is within π/6 of the injected phase. Completeness contours are then determined by computing a two- dimensional moving average of the recovery rate over the sampled Mp sin i and semi-major axis parameter space. An example completeness plot of the RV standard star, GJ 281 is shown in Figure 4.12. This completeness estimate was generated with 1000 simulations, sampling orbital periods between 2 and 20,000 days and RV semi-amplitudes between 50 to 5000 m s−1, and assuming a circular orbit, which is expected if companions migrated to close-in orbits via disk migration (Lin et al. 1996). The FAP was set to 1%. The resulting completeness contour plot indicates a sensitivity to planetary companions of mass 0.3 MJ or higher, out to ∼2 au, and brown dwarf companions out to ∼5 au, with decreasing sensitivity at increasing separations.

Overall, the planet search and detection limits provided by RVSearch indicate that the detection threshold is more conservative than my periodogram search, despite setting the same FAP detection thresholds for both methods. The FAP estimates from RVSearch are typically higher, and there is no consistent offset between the FAP estimates generated by the two methods. This could be the result of the different techniques for FAP determination, where the empirical method employed by RVSearch better accounts for correlated RV noise in the distribution of periodogram peaks.

4.2 Results and Discussion

4.2.1 Individual System Results

The ten RV targets were chosen based on the presence of possible companion-induced RV variability in either infrared or optical data, or to survey possible targets to look for star- planet-disk interactions in the K2 photometry. The stellar mass and age of each star was collected from the literature to characterize a companion mass (see Equation 1.2 in Section

160 Figure 4.12 Example completeness contour plot showing the results of the injection/recovery tests for the RV standard star, GJ 281. The blue points represent simulated planet injections that were successfully recovered and the red points represent planets that were not recovered. The contours indicate different detection probabilities based on the fraction of recovered signals. The simulated planet signal, which was recovered in the periodogram analysis (Figure 4.7), is indicated by the black circle.

161 1.2.3) and to constrain a timescale of planet formation and evolution. Table 4.1 lists the stellar properties of the RV targets, and their sources from the literature. These values were primarily determined from theoretical isochrones based on stellar evolution models.

Stellar masses and ages of young stars can be estimated by comparing their positions on the

H-R diagram relative to isochrone tracks. Observed properties, like luminosity and effective temperature, can then be used to estimate unknown stellar properties such as mass and age

(Cohen & Kuhi 1979). However, model isochrones for pre-main sequence stars are difficult to calibrate for stars with ages <10 Myr (Simon et al. 2017). Luminosity and temperature can be difficult to measure because of disk accretion or magnetic activity (Gully-Santiago et al. 2017; Flores et al. 2019). Uncertainties on stellar masses and ages can be as high as

20% for pre-main sequence stars (Hillenbrand & White 2004; Soderblom 2010).

The ten RV targets were observed intensively over 3 observing season (CI Tau, DK Tau,

V830 Tau, V1075 Tau), or primarily over the final follow-up season (AA Tau, DM Tau, GI

Tau, GM Tau, IQ Tau, LkCa 15). Table 4.1 lists the number of RV observations of each target and the median total RV uncertainties, which typically ranges between 100-200 m s−1. A search for companion signals and injection/recovery simulations were implemented using the techniques outlined above. Results of the analysis are presented below for each individual system.

CI Tau

CI Tau is a CTTS with spectral type K7 and an age of 2.0 ± 0.5 Myr (Guilloteau et al.

2014). Using CN observations to identify the rotation of the disk, Guilloteau et al. (2014) estimated a dynamical mass of 0.8 M , while Simon et al. (2019) estimated a dynamical mass of 0.9 M using interferometry. CI Tau hosts a massive protoplanetary disk with an inclination of 45.7◦ ± 1.1◦, as measured from sub-mm continuum emission, and a mass between ∼20-70 MJ (Andrews & Williams 2005; Guilloteau et al. 2011; McClure et al. 2013; Mohanty et al. 2013).

162 Table 4.1. RV Target Properties and Observations

Target Mass Age N σRV Mass Ref. Age Ref. −1 (M ) (Myr) (m s )

AA Tau 0.7-0.8 2 34 110 1,2 3 CI Tau 0.8-0.9 2 74 110 2,3 4 DK Tau 0.6-0.7 1 54 140 1,5,6 7,8 DM Tau 0.6 4-8 15 100 4 4 GI Tau 0.9 1-4 22 120 1 9 GM Tau 0.1 ··· 18 170 10 ··· IQ Tau 0.8 1-3 25 120 4 4 LkCa 15 1.0 1 30 130 11 12 V830 Tau 1.0 2 87 140 13 13 V1075 Tau 0.7 2 93 170 14 14

Note. — Summary of published stellar mass and age estimates for each RV target, number of observations (N), median random uncertainty (σRV ), and stellar property references. The stellar masses and ages were obtained from: 1) Johns-Krull (2007), 2) Simon et al. (2019), 3) Ricci et al. (2010), 4) Guilloteau et al. (2014), 5) Kraus et al. (2011), 6) Simon et al. (2017), 7) Palla & Stahler (2002), 8) Kraus & Hillenbrand (2009), 9) Guo et al. (2018), 10) Hendler et al. (2017), 11) Pi´etuet al. (2007), 12) Kraus et al. (2012), 13) Donati et al. (2015), 14) Walter et al. (1988)

163 The stellar rotation period of CI Tau is uncertain. Photometric monitoring does not show clear periodicity because of intrinsic variability in the system. Johns-Krull et al.

(2016) report a possible period for CI Tau at ∼7 days, while Percy et al. (2010) report a period of ∼14 days. Artemenko et al. (2012) used 320 photometric observations of CI Tau from Grankin et al. (2007), to estimate a period of ∼16 days, however, independent analysis of these observations indicate a high FAP (∼5%) for this result (Johns-Krull et al. 2016).

Biddle et al. (2018) estimated a rotation period of 6.6 days, using K2 photometry, and

Donati et al. (2020) report a 9.0 day rotation period, which is also present in the K2 data

(see below for additional discussion of the K2 photometry results). The rotation period can also be estimated from the v sin i, inclination, and stellar radius (see Section 3.2.5):

2πR sin i P = . (4.1) rot v sin i

Uncertainties in these measured quantities, however, make it difficult to obtain an accurate stellar rotation period. For example, using the v sin i presented in Chapter 3 (12.0 ± 1.8

km s−1) and inclinations measured by Guilloteau et al. (2014) and McClure et al. (2013)

◦ ◦ (46 and 55 , respectively), as well as an estimated stellar radius of 1.41 R (McClure et al. 2013), the calculated stellar rotation period is between 4.3-4.9 days. However, assuming the

stellar radius, 2.0 R , reported by Donati et al. (2020), the rotation period may instead be between 6.9-8.4 days. Assuming the Basri & Batalha (1990) v sin i estimate of 11 km s−1, which is still within the uncertainty of the infrared v sin i measurement, the rotation period is estimated to be between 4.6-9.2 days, depending on the inclination and radius used in the calculation. Given this range of variability, a definite stellar rotation period for CI Tau could not be determined.

CI Tau was chosen as a priority RV target to confirm a hot Jupiter candidate in the system. Johns-Krull et al. (2016) presented the discovery of a planet candidate of mass 11

MJ with an orbital period of 9 days, using both optical and infrared RV data. Periodogram analysis of the optical data revealed a dominant peak at 9.5 days with a FAP of 1.7% (a

164 peak at 7.2 days was also found at the same FAP). Supporting infrared data confirmed a nearby dominant peak at 9.0 days with a FAP of 0.06%. Combining the infrared and optical RV measurements increased the strength of the 9-day period peak, resulting in a

FAP of 0.08%. The combined dataset indicated a Mp sin i of ∼8 MJ. Given the mass of

◦ the central star (0.8 M ; Simon et al. 2019) and inclination of the disk (46 ; Guilloteau et al. 2014), and assuming that the planet candidate and disk are coplanar, the resulting absolute mass is 11-12 MJ. The measured eccentricity was 0.28 ± 0.16. This property was not well-determined given the large level of astrophysical noise in the system.

Other possible causes of RV variation in the CI Tau system were investigated, such

as starspots modulated at the stellar rotation period, an accretion hotspot, or starlight

scattering off the inner wall of the disk. Although the stellar rotation period was unknown,

starspots were ruled out by comparing the semi-amplitudes of the optical and RV data.

When phased to the 9-day period, the amplitudes matched to within their uncertainties

(Figure 2.10 (bottom)). RV variation caused by starspots is likely to affect optical data to

a larger degree than infrared data, so the matched amplitudes indicated that the detection

may be planetary in nature (Mahmud et al. 2011). No evidence of a long-lived, coherent

accretion hotspot was identified in the photometry over the time baseline established by

the RV measurements (∼5 years), and examination of the emission and absorption lines

did not show expected variations in veiling that would result from a significant accretion

hotspot. Structure at the inner wall of the disk, however, may contribute scattered light

that creates RV variability as the result of the orbital motion of the disk. Given the peak

period detected in the RV periodogram analysis, if scattered light were responsible for the

signal, the structure in the disk causing the RV variability would be located well inside the

0.12 au inner wall of the disk reported by McClure et al. (2013), at a distance of 0.08 au.

While the stellar rotation period could not be definitively identified due to inconsistent

estimates measured from photometry, Johns-Krull et al. (2016) concluded that the planet

hypothesis was the most likely explanation for the RV variability detected in the system.

This conclusion was supported by the diverse datasets used in the analysis, collected over

165 1.0

1.2 0.8

0.6 1.0

Power 0.4 24.44

Normalized Flux 9.06 0.8 0.2 6.57 14.27 0.0 820 830 840 850 860 870 880 890 900 0 5 10 15 20 25 30 BJD - 2450000 +2.457e6 Period (d)

Figure 4.13 The CI Tau K2 lightcurve (left) and the corresponding power spectrum analysis (right) from Biddle et al. (2018). Two major peaks were detected at 6.57 and 9.06 days. The 9-day period was attributed to pulsed accretion caused by a planetary companion.

the course of 10 years from five different facilities, and the consistent RV variability at different wavelengths with the same 9-day periodic signal.

To further support this result, Biddle et al. (2018) looked for evidence of star-planet- disk interactions in the CI Tau system using K2 photometry. Given the inclination of the system, the planet candidate is not likely to transit, but may still cause photometric variability by driving pulsed accretion of disk material onto the host star. Moreover, an eccentric planet can cause accretion with a dominant photometric variability frequency at the planet’s orbital frequency (Teyssandier & Lai 2020). Lomb-Scargle periodogram analysis of the K2 CI Tau lightcurve indicated a strong peak corresponding to a 9.0-day period and a weaker period at 6.6 days (Figure 4.13 from Biddle et al. 2018). Using Equation 4.1 and the v sin i and inclination estimates (assuming the disk and stellar spin axes are aligned), the stellar radius was estimated for each of the two periods, then compared to the stellar radius of 1.41 R reported by McClure et al. (2013). The resulting radii were 1.81 and 2.50

R for the 6.6 and 9.0 day periods, respectively. Given the age and spectral type of CI Tau, the higher radius was deemed unrealistic, based on evolution models (Baraffe et al.

2015). Thus, Biddle et al. (2018) concluded that the 6.6-day period is caused by the stellar rotation and the 9.0-day period is caused by planet-disk interactions.

166 Additionally, Flagg et al. (2019) detected CO in the CI Tau system at the 5.7σ significance level, which they identify as originating from the planet atmosphere. By using infrared spectra that sample different phases of the planet’s orbit, the median stellar signal could be subtracted out while the planet signal remains, allowing for a direct detection of the planet’s spectrum. The CO signature was detected at a velocity amplitude that corresponds to a model-independent companion mass between ∼9-15 MJ.

Despite this additional evidence, Donati et al. (2020) claim that the 9-day period instead originates from stellar activity, and therefore characterizes the rotation period of CI Tau.

Their optical spectropolarimetric observations show RV fluctuations at a similar amplitude as those reported by Johns-Krull et al. (2016), which suggests the detected RV variability in both datasets share the same origin. However, line bisector analysis provides evidence that the RV variations are correlated with the changes in the bisector spans, thus indicating that starspots are likely the cause of the 9-day period (see Section 2.3 for additional details on this technique). The 6.6-day period identified by Biddle et al. (2018) as the stellar rotation period, is not recovered in the spectropolarimetric data. Donati et al. (2020) also calculated a higher radius estimate for CI Tau (2 R ), based on pre-main sequence evolutionary models (Siess et al. 2000). This radius more closely matches the orbital period

of 9 days (Biddle et al. 2018). The authors further posit that the CO signature detected

by Flagg et al. (2019) may instead originate from a structure within the inner disk where

CO is present. While this study refutes the existence of a hot Jupiter in a 9-day orbit,

the system may host other substellar companions. In fact, high-resolution mm-continuum

imaging of the protoplanetary disk in the CI Tau system using ALMA has revealed gaps in

the disk structure beyond 10-100 au, which were likely generated by massive planets in the

early stages of formation (Clarke et al. 2018; Konishi et al. 2018).

The observations obtained in this survey provide the opportunity to shed further light on

the disputed nature of a possible planetary companion to CI Tau using additional follow-up

RVs. The CI Tau dataset has a total of 74 IGRINS observations collected from September

2014 to March 2019, with 15 observations from McDonald Observatory and 59 observations

167 Figure 4.14 CI Tau RVs vs. JD. CI Tau was observed with IGRINS at McDonald Observatory prior to 2016 (orange), and in three separate observing seasons at LDT between 2016 and 2019 (blue).

from LDT. The McDonald IGRINS data was also used in the CI Tau b discovery paper

(Johns-Krull et al. 2016), along with additional infrared and optical spectra. Table A.2 in

Appendix A lists the CI Tau observation times (JDs), RVs, and uncertainties. The CI Tau

RVs are presented in Figure 4.14 as a function of observing time.

Periodogram analysis of the CI Tau RVs does not reveal any signals that meet the 1%

FAP threshold for a statistically significant detection. Excess power at 9 days is present above the noise in the CI Tau IGRINS power spectrum (as indicated by the red dotted line in the top panel of Figure 4.15), but the signal is not significant. Generally, including more data increases the significance of detections, however, repeating the CI Tau periodogram

168 analysis with only the IGRINS/LDT dataset (blue points in Figure 4.14) resulted in a peak at 9 days with a higher power than the full dataset (bottom panel of Figure 4.15). The

IGRINS/LDT power spectrum reveals a period at 9.01 days with a ∼6% FAP. While not

a statistically significant detection, this period matches what has been detected in the CI

Tau data in previous analysis of both RVs and space-based photometry (Johns-Krull et al.

2016; Biddle et al. 2018; Donati et al. 2020).

Given the expected infrared amplitude of 1.08 km s−1 (Johns-Krull et al. 2016) and

the typical RV scatter of 0.55 km s−1 in the IGRINS RV data, the non-detection of the

CI Tau b is surprising. To confirm this, I used RVSearch to perform an injection/recovery

test where 1000 simulated planets were injected into the CI Tau RVs in order to define

the detection limits of this dataset (Howard & Fulton 2016). The detection limits can be

used to estimate the probability that the expected candidate signal, corresponding to an

11 MJ planet in a 9-day orbit, would be detected in the RV data. Figure 4.16 shows the completeness contour plot for the CI Tau RVs at a 1% FAP threshold. At the 1% FAP

threshold, there is an 85.5% probability of detection, assuming the minimum mass of this

candidate (∼8 MJ), and a 93.4% probability, assuming the most probable mass (∼11 MJ). Even at the more conservative 0.1% FAP threshold, there is a 70.5% and 89.5% chance

of detection at the minimum mass and probable mass, respectively. As indicated by the

injection/recovery completeness contour plot, and considering a 50% detection probability

or higher, this dataset is generally sensitive to the full brown dwarf regime up to 0.5 au, and

brown dwarfs at increasingly higher mass out to 1.4 au. The lowest mass planet that can

theoretically be detected is a ∼4 MJ planet, within 0.03 au. At the warm Jupiter regime

boundary (0.1 au), planetary objects with a mass of ∼5 MJ may also be detected. The dataset is also sensitive to higher mass gas giant planets out to 0.5 au (Table 4.2). A planet candidate with a mass of 11 MJ and a 9-day orbital period is estimated to have a detection probability of ∼85-95%.

This analysis shows that there is a high detection probability for a hot Jupiter with the

measured properties presented in Johns-Krull et al. (2016), but the IGRINS RVs show no

169 Figure 4.15 Periodogram analysis of the full CI Tau dataset (top) and IGRINS/LDT data only (bottom). The FAP threshold for a statistically significant detection is shown as a horizontal dashed line. The location of the 9-day period detected by Johns-Krull et al. (2016), Biddle et al. (2018), and Donati et al. (2020) is represented by the vertical dotted red line and additional possible stellar rotation periods are indicated by the dotted orange lines. The power at the 9-day period is increased in the LDT-only dataset (bottom), but does not indicate a statistically significant detection.

170 Figure 4.16 Completeness contour plot illustrating the results of the injection/recovery simulations for CI Tau with a 1% FAP threshold. The simulation injected 1000 fake planetary signals into the dataset; recovered signals are shown in blue and signals that were not recovered are shown in red. The red shaded contours indicate the probability of detection for a given Mp sin i (MJ) and semi-major axis (au). The brown line represents the low-mass star/brown dwarf boundary, the dashed yellow line represents the brown dwarf/planet boundary, and the cyan line is drawn at a mass of 1 MJ. The black circle shows the estimated mass and distance of the hot Jupiter candidate, CI Tau b, assuming the most probable mass of 11 MJ. There is a ∼85-95% probability that this planet candidate would be detected in the dataset at the FAP threshold of 1%, and a ∼70-90% probability of detection at the more conservative 0.1% FAP threshold.

171 evidence of this companion. Johns-Krull et al. (2016) measured a RV semi-amplitude of

0.950 ± 0.207 km s−1 in the combined infrared and optical data, and 1.084 ± 0.250 km s−1 in the infrared data alone. This suggests that the RV amplitudes at both wavelength regimes approximately match, indicating a planetary origin for the 9-day signal. The 9-day period was detected with a FAP of 0.1% in the infrared data from multiple instruments.

A 9-day signal was also detected using K2 photometry, with a FAP of <0.0001% (Biddle

et al. 2018) and in optical RV data, with a FAP on the order of 0.1% (Donati et al.

2020). However, Donati et al. (2020) attributed this signal to stellar rotation rather than a

companion orbital period. Overall, evidence from the literature strongly suggests that there

is a periodic signal at 9 days originating in the CI Tau system, yet the source of that signal

remains unclear. The IGRINS data shows excess power at 9 days, but this peak is not

statistically significant. One option for the lack of a clear detection in this survey is time

domain variability, which could be caused by different levels of stellar activity at different

times. If the signal is activity-induced, it is expected that the corresponding amplitude in

infrared RVs would be lower than the optical, and may vary over time. Given the highly-

active nature of CI Tau, it is also expected that optical instruments would detect significant

RV variation modulated at the rotation period of the star, thus leading to a statistically

significant detection. Additional study of this system is needed to conclude the source of

the 9-day periodic signal.

V830 Tau

V830 Tau is a ∼2 Myr old, WTTS with a mass of 1 M (Donati et al. 2015) and a stellar rotation period of 2.74 days (Xiao et al. 2012; Grankin 2013). This target was chosen for the

RV survey based on the discovery of a 0.8 MJ hot Jupiter candidate in 4.93-day orbit using optical spectroscopy combined with Zeeman Doppler Imaging, an imaging technique used

to map stellar activity features on the surface of a star (Donati et al. 2015, 2016). Donati

et al. (2017) further modeled the stellar activity jitter using a Gaussian process regression

method (see Section 4.1.5) to confirm the companion detection at a FAP of 10−5. Damasso

172 et al. (2020) recently combined photometry, analysis of activity indicators to characterize chromospheric activity, and RVs analyzed with Gaussian process regression, to confirm this detection. The planet signal was not identified and injection/recovery simulations indicated that the planet could have been reliably detected if the RV semi-amplitude of the injected signal is larger than the scatter in the RV residuals (after removing the stellar activity signals). The estimated semi-amplitude of the RV variation associated with the V830 Tau b detection is 60-80 m s−1 (Donati et al. 2016, 2017). Given the IGRINS survey precision

(∼70-120 m s−1), V830 Tau b cannot be confirmed or refuted with the IGRINS RV data.

However, limits on non-detections can be placed on the system.

The V830 Tau dataset includes a total of 87 IGRINS observations, with 11 from

McDonald Observatory, and 76 from LDT. The RVs are listed in Table A.3 in Appendix

A and are shown as a function of time in Figure 4.17 (top). The periodogram analysis indicates a statistically significant peak corresponding to the stellar rotation period at 2.74 days, with a FAP of 0.2% (Figure 4.17 (middle)). This rotation period is consistent with the values estimated in the literature (Xiao et al. 2012; Grankin 2013) and the period identified in the K2 data (Figure 4.18). The RVs phased to this period are also shown in Figure

4.17 (bottom). No additional statistically significant peaks were identified in the system

and additional analysis showed that, while there are minor peaks caused by aliasing of the

2.74-day period in the data, the main peak is not produced by the spectral window function.

This result illustrates a few key points: 1) the V830 Tau stellar rotation period was recovered with a FAP of 0.2% (corresponding to a >3-σ confidence level), further confirming

the accuracy of the RV measurements and method, 2) the activity-induced signal indicates

that V830 Tau has a significant level of magnetic activity, which is expected for a pre-main

sequence star with relatively fast rotation (v sin i ∼30 km s−1; see Chapter 3), 3) while

infrared light is less susceptible to RV variability caused by starspots, it is not immune

to activity features, so additional methods must also be applied to confirm the planetary

nature of any signal detected from infrared RVs.

173 Figure 4.17 RVs vs. time (top), power spectrum (middle), and phased RVs (bottom) for V830 Tau. The dominant peak at 2.74 days is a statistically significant detection, and corresponds to the stellar rotation period (orange dotted vertical line). Expected locations of harmonics are indicated by gray dashed vertical lines and the FAP threshold for a statistically significant detection is shown as a horizontal dashed line. A sinusoidal curve is fit to the phased RVs (orange line) and binned RVs are shown in red. 174 1.20 1.0 2.74 1.15 0.8 1.10 0.6 1.05

Power 0.4 1.00

Normalized Flux 0.2 0.95

0.0 820 830 840 850 860 870 880 890 900 0 5 10 15 20 25 30 BJD - 2450000 +2.457e6 Period (d)

Figure 4.18 The V830 Tau K2 lightcurve (left) and power spectrum (right), analyzed by collaborator Lauren Biddle. The dominant peak at 2.74 days corresponds to the stellar rotation period of the star.

175 Detection limits were defined for the V830 Tau data using RVSearch. Prior to running the injection/recovery simulations, the dominant peak due to stellar rotation was removed by fitting a sinusoid and subtracting it from the RVs. RVSearch typically injects planet signals into the full dataset, then simultaneously fits for known planet detections and searches for additional planets. However, given the more conservative FAP estimate utilized by RVSearch, the peak was not detected at a high enough significance to follow these procedures. Instead, the dominant periodic signal was removed to avoid any significant influence on the synthetic planet recovery outcomes. Figure 4.19 shows the results of this simulation. As expected, V830 Tau b was not identified in this dataset. The gray circle in Figure 4.19 indicates where V830 Tau b lies in Mp sin i and semi-major axis parameter space. The associated detection probability is essentially zero in this region. The lowest

mass of a planet that could theoretically be detected in the V830 Tau data, assuming a

detection probability of at least 50%, is ∼2 MJ at a separation of 0.03 au. Warm Jupiters with a mass >3 MJ may also be detected. Objects in the planetary regime may be identified out to 0.8 au, and objects in the brown dwarf regime out to 2.3 au (Table 4.2).

DK Tau

DK Tau is a ∼1 Myr old (Palla & Stahler 2002; Kraus & Hillenbrand 2009) wide binary system, comprised of two CTTSs, and an angular separation of 2.3000 (White & Ghez 2001) and a physical separation of 340 au (Kraus et al. 2012). The primary star was observed as part of the RV survey and has a mass of ∼0.6-0.7 M (Johns-Krull 2007; Kraus et al. 2011; Simon et al. 2017). Pre-main sequence binaries offer an interesting glimpse into early system evolution. For instance, while the disk dissipation timescale in young binary systems is typically shorter than single stars, continuing accretion has been observed suggesting replenishment of the inner disk. This effect seems to preferentially occur around the primary star and may be enabled by a circumbinary reservoir of material (White & Ghez 2001).

The stellar rotation period estimates from the literature are not consistent, which is likely the result of quasiperiodic variation in the system caused by disk accretion that changes in

176 Figure 4.19 Completeness contour plot illustrating the results of the injection/recovery simulations for V830 Tau (see also Figure 4.16). The gray circle shows the estimated mass and distance of V830 Tau b. As expected given the sensitivity of this survey, there is a very low detection probability for this companion.

177 1.0 2.0 0.8 1.5 0.6

1.0 Power 0.4

Normalized Flux 0.5 0.2 4.28 7.61 15.64 0.0 820 830 840 850 860 870 880 890 900 0 5 10 15 20 25 30 BJD - 2450000 +2.457e6 Period (d)

Figure 4.20 The DK Tau K2 lightcurve (left) and power spectrum (right), analyzed by collaborator Lauren Biddle. Neither the space-based photometry nor the ground-based photometry (provided by collaborator, Brian Skiff) show consistent periodicity in the DK Tau system.

time and location (Herbst et al. 2007). DK Tau is known to be actively accreting material from the disk (Najita et al. 2015), leading to significant dimming events (Rodriguez et al.

2017). Based on photometric variability, Bouvier et al. (1995) and Percy et al. (2010) identified a stellar rotation period of 8.4 and 8.18 days, respectively. Optical RV variability, however, suggests a lower rotation period (4.4 days; Crockett 2011). Xiao et al. (2012) also found evidence of a lower rotation period (4.094 days) based on photometric variability.

The Lowell Observatory ground-based observations (collected and analyzed by collaborator

Brian Skiff), and the K2 data (analyzed by collaborator Lauren Biddle; Figure 4.20), do not show conclusive evidence to support either rotation period.

The motivation for including DK Tau in the RV follow-up came from prior optical and infrared RV analysis. Crockett et al. (2012) calculated the ratio of the RV amplitudes in both wavelength regimes to test if periodic signals detected in the data are consistent with planetary orbital motion (matched amplitudes) or activity-induced RV variability (larger optical amplitudes; see Mahmud et al. (2011) for additional details). The ratio is 1.5, and therefore indicates a larger amplitude in the optical, however, a simple spot model is insufficient to explain the observed features of the combined datasets. The strongest peak in the optical power spectrum corresponded to 4.83 days, which was later updated to 4.4 days with additional data (Crockett 2011). This result contained significant scatter and

178 did not match the 14.93-day period identified in the infrared data. The infrared RVs did not phase well to the optical period, nor did the optical RVs phase well to the infrared period. Therefore, a different physical mechanism may be responsible for the optical and infrared RV modulation. A similar result was found by comparing the CI Tau RVs in both wavelength regimes, which prompted additional observations and analysis leading to the hot Jupiter candidate discovery (Johns-Krull et al. 2016).

Figure 4.21 shows the DK Tau infrared RVs as a function of time (top). A total of

54 IGRINS observations (with only one observation from McDonald Observatory) were included in the dataset. The DK Tau RVs are listed in Table A.4 in Appendix A. A dominant peak at 8.17 days was detected in the power spectrum with a FAP of 0.1% (Figure

4.21 (middle)). This period corresponds to the possible stellar rotation period as identified by Percy et al. (2010), and is 2 times the possible stellar rotation period as identified by

Xiao et al. (2012). Given the similarity between the photometrically-derived stellar rotation period and the RV variability, I conclude that this detection likely signifies activity-induced variability modulated at the rotation period of the star rather than a substellar companion candidate detection. The binned and total RVs, phased to the 8.17-day period, with the sinusoidal fit overlaid, are shown in Figure 4.21 (bottom). No additional significant periodic signals were identified in the RV residuals. This result further illustrates that infrared light is still sensitive to activity-induced RV variability and, in some cases, can confirm the stellar rotation period of a pre-main sequence star.

A closer inspection of the power spectrum shows that the dominant peak at 8.17 days is part of a doublet, with an additional peak of lesser power at 8.33 days (Figure 4.22 (top)).

Doublets are often caused by aliasing due to incomplete sampling of the data, with additional power added to each true peak by the window function. The window function can therefore be used to determine which of the two peaks corresponds to the true periodicity in the RV dataset (Dawson & Fabrycky 2010). Noise in the data prevents characterization of aliases with matched amplitudes and patterns in the window function, therefore, the true period cannot be conclusively determined. However, the 8.17-day period more closely matches

179 Figure 4.21 RVs vs. time (top), power spectrum (middle), and phased RVs (bottom) for DK Tau, as described in Figure 4.17. The dominant peak at 8.17 days is a statistically significant detection, and likely corresponds to the stellar rotation period, which is represented by the orange dotted vertical line.

180 the photometrically-derived stellar rotation periods. Removing the dominant period also removes the alias (Figure 4.22 (bottom)).

The dominant signal in the DK Tau RVs was removed by subtracting the best fit sine curve from the RV data prior to running the injection/recovery simulations using RVSearch.

However, the signal is non-sinusoidal, as indicated by the RV phase curve, thus there may be some residual RV variability caused by stellar rotation that will reduce the detection limits. The completeness contour plot is shown in Figure 4.23. Considering a detection probability of 50% or higher, this dataset is not very sensitive to planetary companions.

The lowest mass planet that can be detected at a distance of 0.03 au is 8 MJ. Only brown dwarfs can be detected within the warm Jupiter regime (>0.1 au), and only low-mass stellar

companions beyond 0.8 au (Table 4.2).

V1075 Tau

V1075 Tau is a WTTS with an estimated age of ∼1.6 Myr and a mass of 0.7 M (Walter et al. 1988). The stellar rotation period was identified as 2.4 days using the Lowell

Observatory ground-based photometry from collaborator, Brian Skiff, and confirmed by

the K2 photometry provided by collaborator, Lauren Biddle (Figure 4.24). This target was

initially chosen somewhat arbitrarily from the full target list based on its relative brightness

(K<9) and the possibility of less RV scatter and a well-characterized rotation period as a

WTTS. V1075 Tau was investigated as an original target of interest with no associated

published results suggesting a substellar candidate in the system. The V1075 Tau dataset

includes a total of 93 observations, with one of those observations being from McDonald

Observatory and the rest from LDT. Preliminary infrared RV observations showed RV

variability that may be attributed to a companion orbital period, however, the phase of

the periodic signal was not initially fully-sampled, and the detection was not statistically

significant. Additional follow-up observations do not support a candidate scenario. A

plot illustrating RVs vs. time and the periodogram analysis is shown in Figure 4.25. No

181 Figure 4.22 The DK Tau power spectrum reveals a doublet with peaks at 8.17 days and 8.33 days (top). Various stellar rotation periods from the literature are indicated by vertical dotted orange lines, and an expected harmonic of the dominant peak is represented as a vertical dotted gray line. Doublets are typically caused by incomplete sampling of the data. Removing the dominant peak also removes the alias (bottom). The window function contains too much noise to conclusively determine which of the two peaks is related to the stellar rotation period, however, the dominant 8.17-day peak more closely matches the photometric estimates in the literature. 182 Figure 4.23 Completeness contour plot illustrating the results of the injection/recovery simulations for DK Tau (see also Figure 4.16).

183 1.0 2.43 1.10 0.8

1.05 0.6

1.00 Power 0.4

Normalized Flux 0.95 0.2

0.0 820 830 840 850 860 870 880 890 900 0 5 10 15 20 25 30 BJD - 2450000 +2.457e6 Period (d)

Figure 4.24 The V1075 Tau K2 lightcurve (left) and power spectrum (right), analyzed by collaborator Lauren Biddle. This result indicates a stellar rotation period at 2.4 days, which is confirmed in the ground-based photometry provided by collaborator Brian Skiff.

significant peaks were recovered in the data. The V1075 Tau RVs are listed in Table A.5 in

Appendix A.

The completeness contour plot for V1075 Tau is shown in Figure 4.26. Injection/recovery simulations with RVSearch indicate sensitivity to companions with a mass up to ∼2 MJ (at

0.03 au) and warm Jupiters up to ∼4 MJ (at 0.1 au). Planetary companions theoretically could have been detected within 0.5 au, and brown dwarfs within 1.5 au (Table 4.2).

K2 Targets

The K2 targets are all CTTSs and were targeted as part of a collaboration with Lauren

Biddle to detect planet candidates using infrared RVs, and characterize star-planet-disk

interactions using K2 photometry. Possible evidence of star-planet-disk interactions,

namely flux caused by disk accretion onto the star modulated by the orbital period of

a companion, can be identified in the K2 power spectra. Biddle et al. (2018) includes additional details of this analysis, and presents K2 photometry indicating possible pulsed

accretion identified in the CI Tau system. The K2 targets include AA Tau, DM Tau, GI

Tau, GM Tau, IQ Tau, and LkCa 15. These targets have all been identified as dipper stars (Rodriguez et al. 2017), which are pre-main sequence stars with an age <10 Myr old

that dim in an unpredictable manner due to transiting dust (or other objects). The large-

amplitude dimmings may be related to planet formation (Cody & Hillenbrand 2010; Cody

184 Figure 4.25 RVs vs. time (top) and power spectrum (bottom) for V1075 Tau. The expected stellar rotation period is indicated by the orange dotted vertical line. The FAP threshold for a statistically significant detection is shown as a horizontal dashed line. There were no significant signals detected in the V1075 Tau RV data.

185 Figure 4.26 Completeness contour plot illustrating the results of the injection/recovery simulations for V1075 Tau (see also Figure 4.16).

186 et al. 2014) and could be caused by exocomets driven to close-approach orbits by massive exoplanets (Ansdell 2018).

AA Tau is a 2.4 Myr old (Ricci et al. 2010) star with a mass of 0.7-0.8 M (Johns- Krull 2007; Simon et al. 2019). The chaotic nature of its lightcurve makes it difficult to

conclusively attribute any periodic behavior to the stellar rotation. DM Tau is a 0.6 M star with an age between 3.5-8 Myr old (Guilloteau et al. 2014) and a rotation period of

15.2 days (Artemenko et al. 2012), which is a longer rotation period than the typical ∼1

week of most low-mass pre-main sequence stars. GI Tau has a mass of 0.9 M (Johns- Krull 2007), an age between 1-4 Myr old (Guo et al. 2018), and a stellar rotation period

of 7.1 days (as determined by collaborator Brian Skiff using ground-based photometry; see

Section 3.2.4). GM Tau has a significantly lower mass than the other RV targets, near the

brown dwarf limit at 0.1 M (Hendler et al. 2017). The K2 data indicate a stellar rotation

period of 2.67 days (Figure 4.27). IQ Tau, a 1-3 Myr old 0.8 M (Guilloteau et al. 2014) star, has a possible rotation period at 6.90 days (Xiao et al. 2012), which is similar to a

peak in the K2 power spectrum at 6.65 days (Figure 4.28). However, this system displays

no periodicity over many years of ground-based photometric monitoring, as conducted by

collaborator Brian Skiff. The final K2 target is LkCa 15, a 1 M (Pi´etuet al. 2007) star with an age possibly as low as 1 Myr old (Kraus & Ireland 2012), and a stellar rotation

period of 5.7 days (Donati et al. 2019). High-resolution imaging of the LkCa 15 system has

revealed evidence of multiple companions in a forming planetary system (Kraus & Ireland

2012; see Section 1.3.3). The K2 data indicates similar periodicity at 5.78 days (Figure

4.29).

The IGRINS RV dataset includes 34 observations of AA Tau, 15 observations of DM

Tau, 22 observations of GI Tau, 18 observations of GM Tau, 25 observations of IQ Tau, and

30 observations of LkCa 15. The K2 targets were observed primarily during the follow-up

season (2018-2019), and therefore have fewer observations overall. Most of the data have

lower than typical resolution, due to the degraded resolution in the K band data during the

final IGRINS/LDT observing season (see Section 2.2.2).

187 1.0

1.5 0.8

0.6 2.67 1.0

Power 0.4

Normalized Flux 0.5 0.2

0.0 820 830 840 850 860 870 880 890 900 0 5 10 15 20 25 30 BJD - 2450000 +2.457e6 Period (d)

Figure 4.27 The GM Tau K2 lightcurve (left) and power spectrum (right), analyzed by collaborator Lauren Biddle. This result indicates a stellar rotation period at 2.67 days.

1.0 2.5 0.8 2.0 0.6 1.5

Power 0.4 1.0 6.65 Normalized Flux 0.2 22.87 0.5 0.0 820 830 840 850 860 870 880 890 900 0 5 10 15 20 25 30 BJD - 2450000 +2.457e6 Period (d)

Figure 4.28 The IQ Tau K2 lightcurve (left) and power spectrum (right), analyzed by collaborator Lauren Biddle. This analysis indicates a stellar rotation period at 6.65 days, similar to the 6.9 day period reported by Xiao et al. (2012).

1.4 1.0

1.2 0.8

1.0 0.6

0.8 Power 0.4 5.78

Normalized Flux 0.6 0.2 20.66

0.4 0.0 820 830 840 850 860 870 880 890 900 0 5 10 15 20 25 30 BJD - 2450000 +2.457e6 Period (d)

Figure 4.29 The LkCa 15 Tau K2 lightcurve (left) and power spectrum (right), analyzed by collaborator Lauren Biddle. This analysis indicates a stellar rotation period at 5.78 days, which confirms the 5.7 day period reported by Donati et al. (2019).

188 The RVs of each target as a function of time and the power spectra generated by the

Lomb-Scargle periodogram analysis are shown in Figure 4.30-4.35. A list of all the individual

RVs are shown in Appendix A. No statistically significant periodic signals were detected in the K2 target RV data. However, an interesting feature of the DM Tau RVs (Figure

4.31 (top)) shows an apparent shift in the relative RVs during the 2018-2019 observing season (JD - 2450000 & 8400), and occurs between December 2018 and January 2019. This shift is not apparent in the other RV datasets, which could suggest that this is a real effect. However, the DM Tau RV scatter is ∼5 times lower than the other RV targets (see

Section 4.2.2 for a detailed discussion of the RV scatter), and this may allow more subtle

RV variability caused by stellar or instrumental effects to be observed in this system. A

similar effect may also be present in the RVs of other targets in the sample, but given the

increased RV scatter presented in those systems, it is not likely that this variability would

be detected. If real, this effect could be evidence of RVs modulated to the stellar rotation

period as starspots on the stellar surface rotate with the star, or it could be an indication

of periodic interactions between the star and disk. However, the K2 data reveal a ∼7.4-day

period and the stellar rotation period was determined to be 15.2 days (Artemenko et al.

2012). If real, the shift occurring within the 79-day gap appears to be related to a longer

term variability that is not resolved given the timeseries of the RV data.

The sensitivity limits, represented by completeness contour plots, are also shown

in Figures 4.36-4.41. Potentially due in part to the limited data and resolution,

there are no significant peaks detected in the periodogram analysis, and, in general,

the completeness contour plots show a decreased sensitivity for these datasets. The

injection/recovery simulations with RVSearch indicate sensitivity to companions (at a 50%

detection probability) with a mass up to ∼2 MJ (DM Tau), ∼5 MJ (GM Tau), ∼7 MJ (LkCa

15), and ∼10 MJ (AA Tau and GI Tau) at 0.03 au. Warm Jupiters may be detected up to

a mass of ∼4 MJ (DM Tau), ∼11 MJ (GM Tau and LkCa 15), and ∼18 MJ (AA Tau and GI Tau). Planetary companions could theoretically be detected within 0.8 au (DM Tau),

0.2 au (GM Tau), 0.1 au (LkCa 15), and 0.05 au (AA Tau and GI Tau). Brown dwarfs may

189 Figure 4.30 RVs vs. time (top) and power spectrum (bottom) for AA Tau (see also Figure 4.25). There were no significant signals detected in the AA Tau RV data.

190 Figure 4.31 RVs vs. time (top) and power spectrum (bottom) for DM Tau (see also Figure 4.25). There were no significant signals detected in the DM Tau RV data.

191 Figure 4.32 RVs vs. time (top) and power spectrum (bottom) for GI Tau (see also Figure 4.25). There were no significant signals detected in the GI Tau RV data.

192 Figure 4.33 RVs vs. time (top) and power spectrum (bottom) for GM Tau (see also Figure 4.25). There were no significant signals detected in the GM Tau RV data.

193 Figure 4.34 RVs vs. time (top) and power spectrum (bottom) for IQ Tau (see also Figure 4.25). There were no significant signals detected in the IQ Tau RV data.

194 Figure 4.35 RVs vs. time (top) and power spectrum (bottom) for LkCa 15 (see also Figure 4.25). There were no significant signals detected in the LkCa 15 RV data.

195 be detected within 1.8 au (DM Tau), 1.1 au (LkCa 15), 1.0 au (GI Tau), 0.9 au (AA Tau) and 0.5 au (GM Tau). The IQ Tau dataset is not sensitive to planetary companions, but can detect brown dwarfs with a mass down to 15 MJ at 0.03 au, and brown dwarfs at the brown dwarf/low-mass star mass boundary at 0.7 au. Table 4.2 provides a summary of the sensitivity limits for each target. In order to fully sample the probabilities for companion detections in the injection/recovery simulations, GM Tau and IQ Tau had to be sampled at periods between 2 and 1000 days with RV semi-amplitudes between 1-7 km s−1 (as opposed

to the 2-20,000 days and 0.05-5 km s−1 used in the other RVSearch simulations). This

change was necessary because of GM Tau’s unusually low mass and the low sensitivity of

the IQ Tau data. The completeness contour plots for GM Tau and IQ Tau given the larger

amplitudes and shorter periods are shown in Figure 4.42.

4.2.2 Discussion

Survey Sensitivity and Occurrence Rate Upper Limits

Based on the RVSearch injection/recovery simulations, and considering a detection

probability of at least 50%, this survey is sensitive to gas giant planets with mass >6

MJ and brown dwarfs within ∼1 au. However, the range of sensitivities varies significantly between targets depending on several factors, such as the number of observations or the level

of RV scatter for each individual system. For instance, at the closest distance sampled (0.03

au), the lowest mass hot Jupiter that can be detected as indicated by injection/recovery

tests is between 2-15 MJ for individual datasets. The lowest mass warm Jupiter that the

survey is sensitive to ranges from a mass of 3-15 MJ. The separation detection limits range from 0.04 au to 0.8 au for companions in the planetary regime, and 0.5 au to 2 au for brown

dwarf companions. Table 4.2 summarizes the detection limits for each individual target

dataset.

A major limiting factor on the survey sensitivity is the RV variation in pre-main sequence

star systems. For comparison, GJ 281, which is a quiet main sequence star of spectral type

M0, demonstrates an estimated sensitivity to ∼0.3 MJ companions at the closest distance

196 Figure 4.36 Completeness contour plot illustrating the results of the injection/recovery simulations for AA Tau (see also Figure 4.16).

197 Figure 4.37 Completeness contour plot illustrating the results of the injection/recovery simulations for DM Tau (see also Figure 4.16).

198 Figure 4.38 Completeness contour plot illustrating the results of the injection/recovery simulations for GI Tau (see also Figure 4.16).

199 Figure 4.39 Completeness contour plot illustrating the results of the injection/recovery simulations for GM Tau (see also Figure 4.16).

200 Figure 4.40 Completeness contour plot illustrating the results of the injection/recovery simulations for IQ Tau (see also Figure 4.16).

201 Figure 4.41 Completeness contour plot illustrating the results of the injection/recovery simulations for LkCa 15 (see also Figure 4.16).

202 Figure 4.42 Completeness contour plots illustrating the results of the injection/recovery simulations for GM Tau and IQ Tau (see also Figure 4.16). The parameters were varied for this simulation to sample shorter periods and larger RV semi-amplitudes.

203 Table 4.2. Completeness of Individual RV Target Datasets

Target HJ Mass WJ Mass Planet a Brown Dwarf a (MJ) (MJ) (au) (au)

AA Tau ≥10 · · · ≤0.05 ≤0.9 CI Tau ≥4 ≥5 ≤0.5 ≤1.4 DK Tau ≥8 · · · ≤0.1 ≤0.8 DM Tau ≥2 ≥4 ≤0.8 ≤1.8 GI Tau ≥10 · · · ≤0.05 ≤1.0 GM Tau ≥5 ≥11 ≤0.2 ≤0.5 IQ Tau · · · · · · · · · ≤0.7 LkCa 15 ≥7 ≥11 ≤0.1 ≤1.1 V830 Tau ≥2 ≥3 ≤0.8 ≤2.3 V1075 Tau ≥2 ≥4 ≤0.5 ≤1.5

Note. — Summary of the survey sensitivity to substellar companions. This table lists the minimum mass of hot Jupiter (HJ) and warm Jupiter (WJ) companions that could be detected given the completeness of each target dataset, as well as the maximum distance at which planetary and brown dwarf companions could be detected.

204 sampled, planetary companions out to ∼2 au, and brown dwarf companions out to ∼5 au

(see Figure 4.12). Theoretically, planets outside the hot/warm Jupiter mass and distance regimes could be detected, as well as brown dwarfs in wider orbits outside the brown dwarf desert.

Limits on occurrence rates based on non-detections provide a basis to constrain planet formation and evolution timescales and processes. Close-in giant planets and brown dwarfs may be more prevalent around pre-main sequence stars if observed before they are destroyed or transformed by evolutionary processes. The results of the RVSearch injection and recovery simulations can be used to define upper limits on the frequency of substellar companions at a given Mp sin i and semi-major axis.

Occurrence rate upper limits were estimated based on the non-detections in the RVs and the sensitivity, as indicated by the injection and recovery simulations. For each of the

10 RV targets, I sampled a grid of various Mp sin i values and semi-major axes. I calculated the survey completeness for detecting simulated planets with the specified properties. I qualified the sensitivity using a detection probability threshold of 99.7%, corresponding to a 3-σ significance level. Assuming the presence of one planet at each of the sampled masses and semi-major axes, the upper limit of the planet occurrence rate is then the number of planets divided by the number of pre-main sequence stars in the sample that were sensitive to a specified companion. This was repeated for each mass and semi-major axis, to provide an estimate of an upper limit on the occurrence rates for companions around pre-main sequence stars. The semi-major axes sampled were 0.03 au, 0.05 au, 0.1 au, 0.2 au, 0.3 au, and 0.4 au, and the mass limits ranged between 20-80 MJ in steps of 10 MJ, and 7-13 MJ

in steps of 1 MJ.

The resulting upper limits on substellar companion occurrence rates as a function of

Mp sin i and semi-major axis are displayed graphically in Figure 4.43. Upper limits on the

occurrence rate of hot Jupiters with a Mp sin i between 7-10 MJ at 0.03 au is 50%, and 25%

for hot Jupiters with a mass ≥10 MJ. At this separation, the upper limit on the occurrence of brown dwarfs is between 10-14%. At 0.05 au, the hot Jupiter upper limit is 50% for a 10

205 MJ companion, and 25% for hot Jupiters of higher mass. The higher mass brown dwarfs

(>20 MJ) are limited to a frequency below 10-11%, and lower mass brown dwarfs have an upper limit of 20%. At the warm Jupiter boundary of 0.1 au, planetary companions have an occurrence rate of <50%, while brown dwarfs have an occurrence rate between <11-25%.

At 0.2 and 0.3 au the survey is only significantly sensitive to brown dwarfs, and upper limits

on brown dwarf frequency are between 14-50%. At 0.4 au, those limits are between 25-50%

for brown dwarfs with a Mp sin i of 50 MJ or higher.

These results apply only to low-mass pre-main sequence stars. The masses of the stars

in the sample are between 0.6 and 1.0 M and the ages are estimated to be ∼1-4 Myr. One

target, GM Tau, has a low mass of 0.1 M , putting it at the boundary between a low-mass

star and brown dwarf. The range of Teff for these targets are approximately between 3000- 5000 K, with a K band magnitude between ∼7-10. Eight of the ten targets are CTTSs and

one is a known binary. All the targets are located in the Taurus-Auriga star forming region.

The upper limits presented here do not account for geometric effects. The orbital

inclination, which measures the angle of a planet’s orbit to the line of sight, can influence

whether a companion is detected using the RV method. For instance, a planet in a face-on

orbit (inclination of 0o) cannot be detected with this technique because the velocity of the

star will not be along the line of sight. This is the reason why a lower limit on the mass

(Mp sin i) is measured instead of the true companion mass. Presumably some companions could be missed if they are inclined with a nearly face-on orbit, even if the survey sensitivity

predicts a high detection probability given the true planet properties. In the case of a low

inclination, if the companion is detected the Mp sin i would be a significant underestimate. However, this effect is small when estimating occurrence rates using the RV method (Wright

et al. 2012). As discussed in Howard & Fulton (2016), orbital orientations are randomly

distributed on the celestial sphere over a uniform distribution of cos i (rather than being

randomly distributed over all angles). The statistical probability of an inclination being

o o between i1 =30 and i2 =90 is P = |cos(i1)-cos(i2)|, which is 87%. Therefore, randomly observing i <30o is far less likely, with a probability of 13%.

206 Figure 4.43 Upper limits placed on occurrence rates of substellar companions with various masses and semi-major axes. The yellow dashed line represents the boundary between the planet and brown dwarf regimes. The survey was not sensitive to low-mass objects at larger separations, as indicated by the gray cross-hatched region. 207 This survey found an upper limit occurrence rate of 25-50% for hot Jupiters with masses above 7 MJ, depending on orbital separation. RV surveys of main sequence stars have estimated a hot Jupiter occurrence rate of ∼1.2% (Marcy et al. 2005; Wright et al. 2012),

1.5% (Cumming et al. 2008), and 0.9%, (Mayor et al. 2011). Grether & Lineweaver (2006)

found that ∼16% of solar-type stars have close companions with an orbital period <5

years, where ∼5% of those companions are giant planets. As discussed in Section 1.3.2,

observational evidence suggests that young hot Jupiter occurrence rates may be higher

than those measured in main sequence systems. Quinn et al. (2014) measured a hot

Jupiter occurrence rate of 1-3% in clusters with ages on the order of 100 Myr (although

the higher frequency could be the result of the host stars being metal-rich). Claimed

hot Jupiter detections in small sample sizes of pre-main sequence stars also suggests a

possible higher occurrence rate (Donati et al. 2017; Yu et al. 2017). Simulations of planet-

disk interactions during disk migration predict a hot Jupiter formation occurrence rate

significantly higher than that observed in main sequence systems (between 8-41% with a

chance of survival during disk migration as high as 15%; Heller 2019). Given an estimated

companion destruction timescale of ∼10 Myr in systems where tidal torques are too weak

to stop disk migration (Heller 2019), it is possible that population statistics vary based on

age. If disk migration is the dominant formation channel for hot Jupiters, this should be

reflected in young companion occurrence rates. On the other hand, alternative migration

mechanisms, such as high-eccentricity migration, occur on timescales longer than the age

of pre-main sequence star systems, and therefore will result in very low occurrence rates.

The limits from this survey are consistent with a higher hot Jupiter frequency in young

systems, and with disk migration as a formation mechanism for hot Jupiters. However,

because this survey presents upper limits, it cannot conclusively rule out occurrence rates

similar to, or lower than main sequence rates, nor can it rule out high-eccentricity migration

as a significant source of hot Jupiters in mature systems.

The upper limit on the occurrence rate of close-in brown dwarfs was estimated to be

between 10-14% for brown dwarfs with masses above 20 MJ at 0.03 au, and up to 20-50%

208 within 0.3 au. Upper limits between 25-50% could only be placed on brown dwarfs out to 0.4 au, with masses above 50 MJ. The existence of the brown dwarf desert suggests that main sequence occurrence rates of close-in brown dwarfs are very low. While Grether

& Lineweaver (2006) found that ∼16% of solar-type stars have close companions with an orbital period <5 years, less than 1% of those companions are brown dwarfs. Grieves et al.

(2017) estimated an occurrence rate of only 0.56% for brown dwarfs around solar-type

stars with periods .300 days. Close-in brown dwarfs within the brown dwarf desert are also predicted to be more prevalent in the early stages of system formation, possibly by

up to an order of magnitude (Armitage & Bonnell 2002). A larger occurrence of brown

dwarfs in close orbits in pre-main sequence systems would indicate that the desert is likely

a feature of evolution rather than formation. In fact, a theoretical interpretation of the

brown dwarf desert boundaries suggests that a lack of brown dwarfs is likely caused by

destructive processes rather than initial conditions (Shahaf & Mazeh 2019). The upper

limits on close-in brown dwarfs is consistent with the theory that the desert is caused by

evolutionary factors, however, as noted with the hot Jupiters, the upper limits also allow

for occurrence rates that are similar to or lower than those in the main sequence.

A larger sample can improve the statistics presented here, and could confirm if the

occurrence rates are higher in young systems. If so, this would support theories of hot

Jupiter disk migration and constrain formation timescales, as well as infer the evolutionary

nature of the brown dwarf desert. Conversely, a lack of young, close-in substellar companions

may suggest that planet formation or migration takes longer than the age of the observed

systems, or that the brown dwarf desert is intrinsic to the formation process of brown dwarfs.

In order to constrain upper limits at a level similar to the main sequence occurrence rates of

short-period massive planets or brown dwarfs, a sample size of over 100 targets is required.

Because of the RV variation exhibited in pre-main sequence star systems, several tens of

observations are typically needed, however, with dense observing cadence this number may

be reduced. Generally, the phase of rotation periods and orbital periods of close-in gas

209 giants and brown dwarfs can be fully sampled with consistent observations over a temporal baseline of roughly one week.

Correlations Between RV Scatter and System Properties

One of the most significant challenges of RV searches for young substellar companions is the high level of RV scatter (or jitter), which can obscure a companion signal in the data. This

RV scatter is largely caused by surface features, like starspots and plages, which manifest as the result of the extreme magnetic activity exhibited by young stars. While observing in the near-infrared regime can reduce starspot-induced RV jitter, it cannot entirely eliminate the effect. Other features in a system may also contribute to the RV scatter, particularly those related to the presence of a circumstellar disk.

The RV survey conducted in this dissertation provides the opportunity to quantify the infrared RV jitter of pre-main sequence stars. I quantified the RV scatter by measuring the standard deviation of the RVs for each individual system. Table 4.3 lists the jitter for each target, as well as potentially related properties, v sin i and classification (CTTS or WTTS).

Figure 4.44 shows a histogram of the measured RV jitter values. The average measured scatter was 550 m s−1. Optical RV scatter measured in systems that are ∼1-3 Myr old is typically ∼1 km s−1 (Mahmud et al. 2011; Crockett et al. 2012; Lagrange et al. 2013;

Brems et al. 2019; Damasso et al. 2020), which can be over an order of magnitude larger than many planet signals. The RV scatter estimated in the infrared is reduced by a factor of ∼2 compared to typical optical RV scatter. This result further motivates the use of RVs measured from infrared spectroscopy over RVs measured at optical wavelengths.

I compared the RV scatter of the IGRINS data to various stellar properties to investigate whether there were any correlations. Additional correlations may provide a better understanding of the RV jitter in various systems, and can direct the sample selection in future surveys.

One notable trend in the RV scatter results is that the CTTSs have a similar level of scatter as the two WTTSs in the sample (V830 Tau and V1075 Tau). This suggests that

210 Figure 4.44 Histogram of the RV scatter for the ten targets in the RV sample. The average measured jitter was 550 m s−1.

211 Table 4.3. Infrared RV Scatter

Target RV Scatter v sin i C/W (m s−1) (km s−1)

AA Tau 590 13.1 C CI Tau 520 12.0 C DK Tau 1100 15.1 C DM Tau 110 6.6 C GI Tau 480 11.3 C GM Tau 610 15.9 C IQ Tau 700 13.5 C LkCa 15 360 14.6 C V830 Tau 540 32.5 W V1075 Tau 550 32.3 W

the jitter is primarily caused by stellar activity, which affects both CTTSs and WTTSs, as opposed to being related primarily to disk phenomena, which would only affect the CTTS systems. However, the WTTSs are also faster rotators than the CTTSs, with a v sin i ∼30 km s−1, compared to the CTTS v sin i values between ∼10-15 km s−1. Stellar rotation and magnetic activity are closely linked, as fast rotation increases stellar activity, and both are known to cause increased RV jitter. Several studies have provided evidence of the relationship between RV scatter and rotation or activity (Saar & Donahue 1997; Bailey et al. 2012; Nguyen et al. 2012; Tal-Or et al. 2018). Stellar activity causes random RV jitter because of asymmetric and changing spot configurations that do not necessarily phase with the stellar rotation period (Mahmud et al. 2011). Fast rotation, in addition to inducing stronger magnetic activity, also broadens spectral lines. This makes RV measurements less precise, which can further increase scatter in the results. The relatively low RV scatter in the DM Tau RVs (∼110 m s−1) supports the correlation between low v sin i and decreased

RV scatter.

Factors related to a disk may also contribute to the RV jitter in the CTTS systems.

Veiling caused by warm dust in the disk can alter spectral lines and affect the precision, thus increasing RV scatter. Crockett et al. (2012) found that cool spot models could not

212 account for the high levels of non-periodic, near-infrared RV variability in their CTTS data, thus suggesting a contribution from the disk. DM Tau and LkCa 15 have both been noted in the literature as having an inner cavity in their disks (Kudo et al. 2018; Alencar et al.

2018), and both display low RV jitter relative to the other targets (100 and 370 m s−1, respectively). If this is a related property, it suggests that features in the disk may be a determining factor in the infrared RV scatter outcomes. Disk accretion may also be a factor. The ground-based photometric observations of IQ Tau, conducted by collaborator

Brian Skiff, indicate that this system has significant, stochastic accretion, as suggested by the lack of peridocity detected in its lightcurve (see Figure 4.45). The higher than average

RV scatter in the IQ Tau RVs (700 m s−1) supports the possible correlation between disk accretion and RV variability. DK Tau also has significantly higher than average RV jitter

(∼1100 m s−1). This result is an outlier, and may be related to its binary status and

misaligned disks (Jensen et al. 2004).

No other evidence of correlations related to intrinsic stellar properties, such as stellar mass or temperature, were identified. Placing the RV targets on an H-R diagram does not reveal any obvious trend relating the RV scatter to evolutionary factors (Figure 4.46).

The correlation of RV jitter with rotation and stellar activity is likely a contributing factor to the measured RV scatter results, and accretion or veiling in the disk may also induce further RV variability. Regardless of these sources of jitter, infrared RVs show significantly decreased scatter relative to optical RVs.

213 Figure 4.45 Photometric observations of IQ Tau from the 2016-2017 observing season, provided by collaborator Brian Skiff. The magnitude is in the V band and HJD refers to the Heliocentric Julian Date. The lightcurve does not show any periodicity, likely as the result of significant accretion, which may contribute to the observed RV jitter.

214 Figure 4.46 A Hertzsprung-Russell diagram with the 9 pre-main sequence stars in the RV sample that have measurements of K magnitudes, parallaxes, and Teff (excluding DK Tau). The RV scatter is indicated by color. Stars from the Pleiades open cluster are overplotted to demonstrate the evolutionary stage of the pre-main sequence star sample relative to the zero age main sequence. There are no obvious trends between RV jitter and position on the diagrams.

215 References

Aigrain, S., Pont, F., & Zucker, S. 2012, MNRAS, 419, 3147

Alencar, S. H. P., Bouvier, J., Donati, J. F., Alecian, E., Folsom, C. P., Grankin, K.,

Hussain, G. A. J., Hill, C., Cody, A. M., Carmona, A., Dougados, C., Gregory, S. G.,

Herczeg, G., M´enard,F., Moutou, C., Malo, L., Takami, M., & Matysse Collaboration.

2018, A&A, 620, A195

Allard, F. & Hauschildt, P. H. 1995, in The Bottom of the Main Sequence - and Beyond,

ed. C. G. Tinney, 32

Andrews, S. M. & Williams, J. P. 2005, ApJ, 631, 1134

Ansdell, M. 2018, in Take a Closer Look, 3

Armitage, P. J. & Bonnell, I. A. 2002, MNRAS, 330, L11

Artemenko, S. A., Grankin, K. N., & Petrov, P. P. 2012, Astronomy Letters, 38, 783

Astropy Collaboration, Price-Whelan, A. M., Sip˝ocz,B. M., G¨unther, H. M., Lim, P. L.,

Crawford, S. M., Conseil, S., Shupe, D. L., Craig, M. W., Dencheva, N., Ginsburg, A.,

Vand erPlas, J. T., Bradley, L. D., P´erez-Su´arez,D., de Val-Borro, M., Aldcroft, T. L.,

Cruz, K. L., Robitaille, T. P., Tollerud, E. J., Ardelean, C., Babej, T., Bach, Y. P.,

Bachetti, M., Bakanov, A. V., Bamford, S. P., Barentsen, G., Barmby, P., Baumbach,

A., Berry, K. L., Biscani, F., Boquien, M., Bostroem, K. A., Bouma, L. G., Brammer,

G. B., Bray, E. M., Breytenbach, H., Buddelmeijer, H., Burke, D. J., Calderone, G.,

216 Cano Rodr´ıguez,J. L., Cara, M., Cardoso, J. V. M., Cheedella, S., Copin, Y., Corrales,

L., Crichton, D., D’Avella, D., Deil, C., Depagne, E.,´ Dietrich, J. P., Donath, A.,

Droettboom, M., Earl, N., Erben, T., Fabbro, S., Ferreira, L. A., Finethy, T., Fox,

R. T., Garrison, L. H., Gibbons, S. L. J., Goldstein, D. A., Gommers, R., Greco, J. P.,

Greenfield, P., Groener, A. M., Grollier, F., Hagen, A., Hirst, P., Homeier, D., Horton,

A. J., Hosseinzadeh, G., Hu, L., Hunkeler, J. S., Ivezi´c, Z.,ˇ Jain, A., Jenness, T., Kanarek,

G., Kendrew, S., Kern, N. S., Kerzendorf, W. E., Khvalko, A., King, J., Kirkby, D.,

Kulkarni, A. M., Kumar, A., Lee, A., Lenz, D., Littlefair, S. P., Ma, Z., Macleod, D. M.,

Mastropietro, M., McCully, C., Montagnac, S., Morris, B. M., Mueller, M., Mumford,

S. J., Muna, D., Murphy, N. A., Nelson, S., Nguyen, G. H., Ninan, J. P., N¨othe, M., Ogaz,

S., Oh, S., Parejko, J. K., Parley, N., Pascual, S., Patil, R., Patil, A. A., Plunkett, A. L.,

Prochaska, J. X., Rastogi, T., Reddy Janga, V., Sabater, J., Sakurikar, P., Seifert, M.,

Sherbert, L. E., Sherwood-Taylor, H., Shih, A. Y., Sick, J., Silbiger, M. T., Singanamalla,

S., Singer, L. P., Sladen, P. H., Sooley, K. A., Sornarajah, S., Streicher, O., Teuben, P.,

Thomas, S. W., Tremblay, G. R., Turner, J. E. H., Terr´on,V., van Kerkwijk, M. H., de

la Vega, A., Watkins, L. L., Weaver, B. A., Whitmore, J. B., Woillez, J., Zabalza, V., &

Astropy Contributors. 2018, AJ, 156, 123

Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., Greenfield, P., Droettboom, M.,

Bray, E., Aldcroft, T., Davis, M., Ginsburg, A., Price-Whelan, A. M., Kerzendorf, W. E.,

Conley, A., Crighton, N., Barbary, K., Muna, D., Ferguson, H., Grollier, F., Parikh,

M. M., Nair, P. H., Unther, H. M., Deil, C., Woillez, J., Conseil, S., Kramer, R., Turner,

J. E. H., Singer, L., Fox, R., Weaver, B. A., Zabalza, V., Edwards, Z. I., Azalee Bostroem,

K., Burke, D. J., Casey, A. R., Crawford, S. M., Dencheva, N., Ely, J., Jenness, T.,

Labrie, K., Lim, P. L., Pierfederici, F., Pontzen, A., Ptak, A., Refsdal, B., Servillat, M.,

& Streicher, O. 2013, A&A, 558, A33

Bailey, John I., I., White, R. J., Blake, C. H., Charbonneau, D., Barman, T. S., Tanner,

A. M., & Torres, G. 2012, ApJ, 749, 16

217 Baraffe, I., Homeier, D., Allard, F., & Chabrier, G. 2015, A&A, 577, A42

Basri, G. & Batalha, C. 1990, ApJ, 363, 654

Bastien, F. A., Stassun, K. G., Pepper, J., Wright, J. T., Aigrain, S., Basri, G., Johnson,

J. A., Howard, A. W., & Walkowicz, L. M. 2014, AJ, 147, 29

Bean, J. L., Seifahrt, A., Hartman, H., Nilsson, H., Wiedemann, G., Reiners, A., Dreizler,

S., & Henry, T. J. 2010, ApJ, 713, 410

Bevington, P. R. & Robinson, D. K. 1992, Data reduction and error analysis for the physical

sciences

Biddle, L. I., Johns-Krull, C. M., Llama, J., Prato, L., & Skiff, B. A. 2018, ApJ, 853, L34

Blake, C. H., Charbonneau, D., & White, R. J. 2010, ApJ, 723, 684

Boisse, I., Bouchy, F., H´ebrard,G., Bonfils, X., Santos, N., & Vauclair, S. 2011, A&A, 528,

A4

Bouvier, J., Covino, E., Kovo, O., Martin, E. L., Matthews, J. M., Terranegra, L., & Beck,

S. C. 1995, A&A, 299, 89

Brems, S. S., K¨urster,M., Trifonov, T., Reffert, S., & Quirrenbach, A. 2019, A&A, 632,

A37

Cale, B., Plavchan, P., LeBrun, D., Gagn´e,J., Gao, P., Tanner, A., Beichman, C., Xuesong

Wang, S., Gaidos, E., Teske, J., Ciardi, D., Vasisht, G., Kane, S. R., & von Braun, K.

2019, AJ, 158, 170

Casagrande, L., Flynn, C., & Bessell, M. 2008, MNRAS, 389, 585

Clarke, C. J., Tazzari, M., Juhasz, A., Rosotti, G., Booth, R., Facchini, S., Ilee, J. D.,

Johns-Krull, C. M., Kama, M., Meru, F., & Prato, L. 2018, ApJ, 866, L6

Cody, A. M. & Hillenbrand, L. A. 2010, ApJS, 191, 389

218 Cody, A. M., Stauffer, J., Baglin, A., Micela, G., Rebull, L. M., Flaccomio, E., Morales-

Calder´on,M., Aigrain, S., Bouvier, J., Hillenbrand, L. A., Gutermuth, R., Song, I.,

Turner, N., Alencar, S. H. P., Zwintz, K., Plavchan, P., Carpenter, J., Findeisen, K.,

Carey, S., Terebey, S., Hartmann, L., Calvet, N., Teixeira, P., Vrba, F. J., Wolk, S.,

Covey, K., Poppenhaeger, K., G¨unther, H. M., Forbrich, J., Whitney, B., Affer, L.,

Herbst, W., Hora, J., Barrado, D., Holtzman, J., Marchis, F., Wood, K., Medeiros

Guimar˜aes,M., Lillo Box, J., Gillen, E., McQuillan, A., Espaillat, C., Allen, L., D’Alessio,

P., & Favata, F. 2014, AJ, 147, 82

Cohen, M. & Kuhi, L. V. 1979, ApJS, 41, 743

Crockett, C. J. 2011, PhD thesis, University of California, Los Angeles

Crockett, C. J., Mahmud, N. I., Prato, L., Johns-Krull, C. M., Jaffe, D. T., & Beichman,

C. A. 2011, ApJ, 735, 78

Crockett, C. J., Mahmud, N. I., Prato, L., Johns-Krull, C. M., Jaffe, D. T., Hartigan, P. M.,

& Beichman, C. A. 2012, ApJ, 761, 164

Cumming, A., Butler, R. P., Marcy, G. W., Vogt, S. S., Wright, J. T., & Fischer, D. A.

2008, PASP, 120, 531

Damasso, M., Bonomo, A. S., Astudillo-Defru, N., Bonfils, X., Malavolta, L., Sozzetti,

A., Lopez, E., Zeng, L., Haywood, R. D., Irwin, J. M., Mortier, A., Vanderburg, A.,

Maldonado, J., Lanza, A. F., Affer, L., Almenara, J. M., Benatti, S., Biazzo, K.,

Bignamini, A., Borsa, F., Bouchy, F., Buchhave, L. A., Cameron, A. C., Carleo, I.,

Charbonneau, D., Claudi, R., Cosentino, R., Covino, E., Delfosse, X., Desidera, S., Di

Fabrizio, L., Dressing, C., Esposito, M., Fares, R., Figueira, P., Fiorenzano, A. F. M.,

Forveille, T., Giacobbe, P., Gonz´alez-Alvarez,´ E., Gratton, R., Harutyunyan, A., Johnson,

J. A., Latham, D. W., Leto, G., Lopez-Morales, M., Lovis, C., Maggio, A., Mancini, L.,

Masiero, S., Mayor, M., Micela, G., Molinari, E., Motalebi, F., Murgas, F., Nascimbeni,

V., Pagano, I., Pepe, F., Phillips, D. F., Piotto, G., Poretti, E., Rainer, M., Rice, K.,

219 Santos, N. C., Sasselov, D., Scandariato, G., S´egransan,D., Smareglia, R., Udry, S.,

Watson, C., & W¨unsche, A. 2018, A&A, 615, A69

Damasso, M., Lanza, A. F., Benatti, S., Rajpaul, V. M., Mallonn, M., Desidera, S.,

Biazzo, K., D’Orazi, V., Malavolta, L., Nardiello, D., Rainer, M., Borsa, F., Affer, L.,

Bignamini, A., Bonomo, A. S., Carleo, I., Claudi, R., Cosentino, R., Covino, E., Giacobbe,

P., Gratton, R., Harutyunyan, A., Knapic, C., Leto, G., Maggio, A., Maldonado, J.,

Mancini, L., Micela, G., Molinari, E., Nascimbeni, V., Pagano, I., Piotto, G., Poretti,

E., Scandariato, G., Sozzetti, A., Capuzzo Dolcetta, R., Di Mauro, M. P., Carosati, D.,

Fiorenzano, A., Frustagli, G., Pedani, M., Pinamonti, M., Stoev, H., & Turrini, D. 2020,

arXiv e-prints, arXiv:2008.09445

Davison, C. L., White, R. J., Henry, T. J., Riedel, A. R., Jao, W. C., Bailey, J. I., I., Quinn,

S. N., Cantrell, J. R., Subasavage, J. P., & Winters, J. G. 2015, AJ, 149, 106

Dawson, R. I. & Fabrycky, D. C. 2010, ApJ, 722, 937

Deming, D., Espenak, F., Jennings, D. E., Brault, J. W., & Wagner, J. 1987, ApJ, 316, 771

Desort, M., Lagrange, A. M., Galland, F., Udry, S., & Mayor, M. 2007, A&A, 473, 983

Donati, J. F., Bouvier, J., Alencar, S. H., Hill, C., Carmona, A., Folsom, C. P., M´enard,F.,

Gregory, S. G., Hussain, G. A., Grankin, K., Moutou, C., Malo, L., Takami, M., Herczeg,

G. J., & MaTYSSE Collaboration. 2019, MNRAS, 483, L1

Donati, J. F., Bouvier, J., Alencar, S. H., Moutou, C., Malo, L., Takami, M., M´enard,F.,

Dougados, C., Hussain, G. A., & The Matysse Collaboration. 2020, MNRAS, 491, 5660

Donati, J. F., H´ebrard,E., Hussain, G. A. J., Moutou, C., Malo, L., Grankin, K., Vidotto,

A. A., Alencar, S. H. P., Gregory, S. G., Jardine, M. M., Herczeg, G., Morin, J., Fares,

R., M´enard,F., Bouvier, J., Delfosse, X., Doyon, R., Takami, M., Figueira, P., Petit, P.,

Boisse, I., & MaTYSSE Collaboration. 2015, MNRAS, 453, 3706

220 Donati, J. F., Moutou, C., Malo, L., Baruteau, C., Yu, L., H´ebrard,E., Hussain, G.,

Alencar, S., M´enard,F., Bouvier, J., Petit, P., Takami, M., Doyon, R., & Cameron,

A. C. 2016, Nature, 534, 662

Donati, J.-F., Yu, L., Moutou, C., Cameron, A. C., Malo, L., Grankin, K., H´ebrard,E.,

Hussain, G. A. J., Vidotto, A. A., Alencar, S. H. P., Haywood, R. D., Bouvier, J., Petit,

P., Takami, M., Herczeg, G. J., Gregory, S. G., Jardine, M. M., & Morin, J. 2017,

MNRAS, 465, 3343

Endl, M., Cochran, W. D., Tull, R. G., & MacQueen, P. J. 2003, AJ, 126, 3099

Figueira, P., Pepe, F., Lovis, C., & Mayor, M. 2010a, A&A, 515, A106

Figueira, P., Pepe, F., Melo, C. H. F., Santos, N. C., Lovis, C., Mayor, M., Queloz, D.,

Smette, A., & Udry, S. 2010b, A&A, 511, A55

Flagg, L., Johns-Krull, C. M., Nofi, L., Llama, J., Prato, L., Sullivan, K., Jaffe, D. T., &

Mace, G. 2019, ApJ, 878, L37

Flores, C., Connelley, M. S., Reipurth, B., & Boogert, A. 2019, ApJ, 882, 75

Fulton, B. J., Petigura, E. A., Blunt, S., & Sinukoff, E. 2018, RadVel: General toolkit for

modeling Radial Velocities

Gaia Collaboration. 2018, VizieR Online Data Catalog, I/345

Grankin, K. N. 2013, Astronomy Letters, 39, 251

Grankin, K. N., Melnikov, S. Y., Bouvier, J., Herbst, W., & Shevchenko, V. S. 2007, A&A,

461, 183

Gray, D. F. 2008, The Observation and Analysis of Stellar Photospheres

Grether, D. & Lineweaver, C. H. 2006, ApJ, 640, 1051

221 Grieves, N., Ge, J., Thomas, N., Ma, B., Sithajan, S., Ghezzi, L., Kimock, B., Willis, K.,

De Lee, N., Lee, B., Fleming, S. W., Agol, E., Troup, N., Paegert, M., Schneider, D. P.,

Stassun, K., Varosi, F., Zhao, B., Jian, L., Li, R., Porto de Mello, G. F., Bizyaev, D.,

Pan, K., Dutra-Ferreira, L., Lorenzo-Oliveira, D., Santiago, B. X., da Costa, L. N., Maia,

M. A. G., Ogando, R. L. C., & del Peloso, E. F. 2017, MNRAS, 467, 4264

Grunblatt, S. K., Howard, A. W., & Haywood, R. D. 2015, ApJ, 808, 127

Guilloteau, S., Dutrey, A., Pi´etu,V., & Boehler, Y. 2011, A&A, 529, A105

Guilloteau, S., Simon, M., Pi´etu,V., Di Folco, E., Dutrey, A., Prato, L., & Chapillon, E.

2014, A&A, 567, A117

Gully-Santiago, M. A., Herczeg, G. J., Czekala, I., Somers, G., Grankin, K., Covey, K. R.,

Donati, J. F., Alencar, S. H. P., Hussain, G. A. J., Shappee, B. J., Mace, G. N., Lee,

J.-J., Holoien, T. W. S., Jose, J., & Liu, C.-F. 2017, ApJ, 836, 200

Guo, Z., Herczeg, G. J., Jose, J., Fu, J., Chiang, P.-S., Grankin, K., Michel, R., Kesh

Yadav, R., Liu, J., Chen, W.-p., Li, G., Xue, H., Niu, H., Subramaniam, A., Sharma, S.,

Prasert, N., Flores-Fajardo, N., Castro, A., & Altamirano, L. 2018, ApJ, 852, 56

Haywood, R. 2016, Stellar Activity as a Source of Radial-Velocity Variability, 13–44

Haywood, R. D., Collier Cameron, A., Queloz, D., Barros, S. C. C., Deleuil, M., Fares, R.,

Gillon, M., Lanza, A. F., Lovis, C., Moutou, C., Pepe, F., Pollacco, D., Santerne, A.,

S´egransan,D., & Unruh, Y. C. 2014, MNRAS, 443, 2517

Heller, R. 2019, A&A, 628, A42

Hendler, N. P., Mulders, G. D., Pascucci, I., Greenwood, A., Kamp, I., Henning, T., M´enard,

F., Dent, W. R. F., & Evans, Neal J., I. 2017, ApJ, 841, 116

Herbst, W., Eisl¨offel,J., Mundt, R., & Scholz, A. 2007, in Protostars and Planets V, ed.

B. Reipurth, D. Jewitt, & K. Keil, 297

222 Hillenbrand, L., Isaacson, H., Marcy, G., Barenfeld, S., Fischer, D., & Howard, A. 2015,

in Cambridge Workshop on Cool Stars, Stellar Systems, and the Sun, Vol. 18, 18th

Cambridge Workshop on Cool Stars, Stellar Systems, and the Sun, 759–766

Hillenbrand, L. A. & White, R. J. 2004, ApJ, 604, 741

Hojjatpanah, S., Oshagh, M., Figueira, P., Santos, N. C., Amazo-G´omez, E. M., Sousa,

S. G., Adibekyan, V., Akinsanmi, B., Demangeon, O., Faria, J., Gomes da Silva, J., &

Meunier, N. 2020, A&A, 639, A35

Howard, A. W. & Fulton, B. J. 2016, PASP, 128, 114401

Huerta, M., Johns-Krull, C. M., Prato, L., Hartigan, P., & Jaffe, D. T. 2008, ApJ, 678, 472

Jensen, E. L. N., Mathieu, R. D., Donar, A. X., & Dullighan, A. 2004, ApJ, 600, 789

Johns-Krull, C. M. 2007, ApJ, 664, 975

Johns-Krull, C. M., McLane, J. N., Prato, L., Crockett, C. J., Jaffe, D. T., Hartigan, P. M.,

Beichman, C. A., Mahmud, N. I., Chen, W., Skiff, B. A., Cauley, P. W., Jones, J. A., &

Mace, G. N. 2016, ApJ, 826, 206

Konishi, M., Hashimoto, J., & Hori, Y. 2018, ApJ, 859, L28

Kosiarek, M. R. & Crossfield, I. J. M. 2020, AJ, 159, 271

Kraus, A. L. & Hillenbrand, L. A. 2009, ApJ, 704, 531

Kraus, A. L. & Ireland, M. J. 2012, ApJ, 745, 5

Kraus, A. L., Ireland, M. J., Hillenbrand, L. A., & Martinache, F. 2012, ApJ, 745, 19

Kraus, A. L., Ireland, M. J., Martinache, F., & Hillenbrand, L. A. 2011, ApJ, 731, 8

Kudo, T., Hashimoto, J., Muto, T., Liu, H. B., Dong, R., Hasegawa, Y., Tsukagoshi, T., &

Konishi, M. 2018, ApJ, 868, L5

223 Kuerster, M., Schmitt, J. H. M. M., Cutispoto, G., & Dennerl, K. 1997, A&A, 320, 831

Lagrange, A. M., Meunier, N., Chauvin, G., Sterzik, M., Galland, F., Lo Curto, G., Rameau,

J., & Sosnowska, D. 2013, A&A, 559, A83

Lin, D. N. C., Bodenheimer, P., & Richardson, D. C. 1996, Nature, 380, 606

Livingston, W. & Wallace, L. 1991, An atlas of the solar spectrum in the infrared from 1850

to 9000 cm-1 (1.1 to 5.4 micrometer)

Lomb, N. R. 1976, Ap&SS, 39, 447

Mahadevan, S. & Ge, J. 2009, ApJ, 692, 1590

Mahmud, N. I., Crockett, C. J., Johns-Krull, C. M., Prato, L., Hartigan, P. M., Jaffe, D. T.,

& Beichman, C. A. 2011, ApJ, 736, 123

Mann, A. W., Gaidos, E., Mace, G. N., Johnson, M. C., Bowler, B. P., LaCourse, D.,

Jacobs, T. L., Vanderburg, A., Kraus, A. L., Kaplan, K. F., & Jaffe, D. T. 2016a, ApJ,

818, 46

Mann, A. W., Newton, E. R., Rizzuto, A. C., Irwin, J., Feiden, G. A., Gaidos, E., Mace,

G. N., Kraus, A. L., James, D. J., Ansdell, M., Charbonneau, D., Covey, K. R., Ireland,

M. J., Jaffe, D. T., Johnson, M. C., Kidder, B., & Vanderburg, A. 2016b, AJ, 152, 61

Marcy, G., Butler, R. P., Fischer, D., Vogt, S., Wright, J. T., Tinney, C. G., & Jones,

H. R. A. 2005, Progress of Theoretical Physics Supplement, 158, 24

Marcy, G. W., Butler, R. P., Williams, E., Bildsten, L., Graham, J. R., Ghez, A. M., &

Jernigan, J. G. 1997, ApJ, 481, 926

Mayor, M., Marmier, M., Lovis, C., Udry, S., S´egransan,D., Pepe, F., Benz, W., Bertaux,

J. ., Bouchy, F., Dumusque, X., Lo Curto, G., Mordasini, C., Queloz, D., & Santos, N. C.

2011, ArXiv e-prints

224 McClure, M. K., Calvet, N., Espaillat, C., Hartmann, L., Hern´andez,J., Ingleby, L.,

Luhman, K. L., D’Alessio, P., & Sargent, B. 2013, ApJ, 769, 73

Mohanty, S., Greaves, J., Mortlock, D., Pascucci, I., Scholz, A., Thompson, M., Apai, D.,

Lodato, G., & Looper, D. 2013, ApJ, 773, 168

Muirhead, P. S., Edelstein, J., Erskine, D. J., Wright, J. T., Muterspaugh, M. W., Covey,

K. R., Wishnow, E. H., Hamren, K., Andelson, P., Kimber, D., Mercer, T., Halverson,

S. P., Vand erburg, A., Mondo, D., Czeszumska, A., & Lloyd, J. P. 2011, PASP, 123, 709

Najita, J. R., Andrews, S. M., & Muzerolle, J. 2015, MNRAS, 450, 3559

Nelder, J. A. & Mead, R. 1965, The Computer Journal, 7, 308

Nguyen, D. C., Brandeker, A., van Kerkwijk, M. H., & Jayawardhana, R. 2012, The

Astrophysical Journal, 745, 119

Oshagh, M., Santos, N. C., Figueira, P., Barros, S. C. C., Donati, J. F., Adibekyan, V.,

Faria, J. P., Watson, C. A., Cegla, H. M., Dumusque, X., H´ebrard,E., Demangeon, O.,

Dreizler, S., Boisse, I., Deleuil, M., Bonfils, X., Pepe, F., & Udry, S. 2017, A&A, 606,

A107

O’Toole, S. J., Tinney, C. G., Jones, H. R. A., Butler, R. P., Marcy, G. W., Carter, B., &

Bailey, J. 2009, MNRAS, 392, 641

Palla, F. & Stahler, S. W. 2002, ApJ, 581, 1194

Percy, J. R., Grynko, S., Seneviratne, R., & Herbst, W. 2010, PASP, 122, 753

Pi´etu,V., Dutrey, A., & Guilloteau, S. 2007, A&A, 467, 163

Piskunov, N. 1999, in Astrophysics and Space Science Library, Vol. 243, Polarization, ed.

K. N. Nagendra & J. O. Stenflo, 515–525

Piskunov, N. E., Kupka, F., Ryabchikova, T. A., Weiss, W. W., & Jeffery, C. S. 1995,

A&AS, 112, 525

225 Quinn, S. N., White, R. J., Latham, D. W., Buchhave, L. A., Torres, G., Stefanik, R. P.,

Berlind, P., Bieryla, A., Calkins, M. C., Esquerdo, G. A., F˝ur´esz,G., Geary, J. C., &

Szentgyorgyi, A. H. 2014, ApJ, 787, 27

Rajpaul, V., Aigrain, S., Osborne, M. A., Reece, S., & Roberts, S. 2015, MNRAS, 452, 2269

Rajpaul, V., Aigrain, S., & Roberts, S. 2016, MNRAS, 456, L6

Reiners, A., Bean, J. L., Huber, K. F., Dreizler, S., Seifahrt, A., & Czesla, S. 2010, ApJ,

710, 432

Ricci, L., Testi, L., Natta, A., Neri, R., Cabrit, S., & Herczeg, G. J. 2010, A&A, 512, A15

Rodriguez, J. E., Ansdell, M., Oelkers, R. J., Cargile, P. A., Gaidos, E., Cody, A. M.,

Stevens, D. J., Somers, G., James, D., Beatty, T. G., Siverd, R. J., Lund, M. B., Kuhn,

R. B., Gaudi, B. S., Pepper, J., & Stassun, K. G. 2017, ApJ, 848, 97

Ryabchikova, T. & Pakhomov, Y. 2015, Baltic Astronomy, 24, 453

Saar, S. H. & Donahue, R. A. 1997, ApJ, 485, 319

Scargle, J. D. 1982, ApJ, 263, 835

Seifahrt, A. & K¨aufl,H. U. 2008, A&A, 491, 929

Shahaf, S. & Mazeh, T. 2019, MNRAS, 487, 3356

Siess, L., Dufour, E., & Forestini, M. 2000, A&A, 358, 593

Simon, M., Guilloteau, S., Beck, T. L., Chapillon, E., Di Folco, E., Dutrey, A., Feiden,

G. A., Grosso, N., Pi´etu,V., Prato, L., & Schaefer, G. H. 2019, ApJ, 884, 42

Simon, M., Guilloteau, S., Di Folco, E., Dutrey, A., Grosso, N., Pi´etu,V., Chapillon, E.,

Prato, L., Schaefer, G. H., Rice, E., & Boehler, Y. 2017, ApJ, 844, 158

Soderblom, D. R. 2010, ARA&A, 48, 581

226 Tal-Or, L., Zechmeister, M., Reiners, A., Jeffers, S. V., Sch¨ofer,P., Quirrenbach, A., Amado,

P. J., Ribas, I., Caballero, J. A., Aceituno, J., Bauer, F. F., B´ejar,V. J. S., Czesla, S.,

Dreizler, S., Fuhrmeister, B., Hatzes, A. P., Johnson, E. N., K¨urster,M., Lafarga, M.,

Montes, D., Morales, J. C., Reffert, S., Sadegi, S., Seifert, W., & Shulyak, D. 2018, A&A,

614, A122

Teyssandier, J. & Lai, D. 2020, MNRAS, 495, 3920

Torres, C. A. O., Quast, G. R., da Silva, L., de La Reza, R., Melo, C. H. F., & Sterzik, M.

2006, A&A, 460, 695

VanderPlas, J. T. 2018, ApJS, 236, 16

Walter, F. M., Brown, A., Mathieu, R. D., Myers, P. C., & Vrba, F. J. 1988, AJ, 96, 297

White, R. J. & Ghez, A. M. 2001, ApJ, 556, 265

Wright, J. T., Marcy, G. W., Howard, A. W., Johnson, J. A., Morton, T. D., & Fischer,

D. A. 2012, ApJ, 753, 160

Xiao, H. Y., Covey, K. R., Rebull, L., Charbonneau, D., Mandushev, G., O’Donovan, F.,

Slesnick, C., & Lloyd, J. P. 2012, ApJS, 202, 7

Yu, L., Donati, J.-F., H´ebrard,E. M., Moutou, C., Malo, L., Grankin, K., Hussain, G.,

Collier Cameron, A., Vidotto, A. A., Baruteau, C., Alencar, S. H. P., Bouvier, J., Petit,

P., Takami, M., Herczeg, G., Gregory, S. G., Jardine, M., Morin, J., M´enard,F., &

Matysse Collaboration. 2017, MNRAS

227 Chapter 5

Conclusions and Summary

This dissertation presents the results of a near-infrared radial velocity survey to find young substellar companions. Observational evidence is needed to better understand planet formation and evolution theory. In particular, the formation mechanism of hot and warm

Jupiters is unclear: whether they primarily migrate on short timescales (typically within a few Myr) via disk migration, or over long timescales (∼1 Gyr) via high-eccentricity migration, or even possibly forming in situ, is currently unknown. It is also unknown if the brown dwarf desert (a dearth of close-in brown dwarfs around solar-type stars) originates from formation or later evolution. Finding these close-in, massive substellar companions during the early stages of formation can place constraints on their formation timescale and provide insights into their formation and evolutionary processes. Limits on the occurrence rates of these objects can also constrain theories and offer a comparison between young and mature systems. Infrared spectroscopy is a promising tool to both characterize stellar properties and identify stellar RV variation caused by the presence of a substellar companion.

Despite the potential impact a RV survey of young stars may have, there are many challenges that must be considered. Pre-main sequence stars exhibit strong magnetic activity with large and numerous starspots. These features also cause RV variation that can mimic a planet signal. By conducting this survey in the infrared regime instead of the optical, the contrast between the starspots and stellar surface is greatly decreased, which also decreases activity-induced RV variability. Starspots are typically modulated at the

228 stellar rotation period, so photometry can be combined with infrared RVs to identify the source of periodic signals. Starspots can also increase scatter in a non-periodic way because of asymmetry and changing configurations over the surface of the star.

A total of 963 infrared spectra for this survey were collected using the high-resolution

(R∼45,000) infrared spectrograph, IGRINS at the Lowell Discovery Telescope. The sample consisted of 70 pre-main sequence stars in the Taurus-Auriga star-forming region, with ages between ∼1-5 Myr, and contained both CTTSs and WTTSs. Stellar properties were characterized for all 70 targets, and a subset of 10 targets were identified for RV follow-up observations.

The infrared spectra enabled stellar properties to be calculated. The v sin i was measured with a standard cross correlation technique, and the Teff was measured from line-depth ratios. Both ground-based and space-based photometry were used to characterize the stellar rotation period, and a limit was placed on the stellar radius by combining the rotation period and v sin i. A comparison of single and multiple stars showed that multiples typically have a higher measured v sin i, which may be caused by contamination from companion spectral lines, shorter disk lifetimes, or tidal interactions in hierarchical triples. A comparison of optical and infrared v sin i estimates showed no significant differences regardless of whether the star has a circumstellar disk or not, which indicates that CO contamination from the disk does not significantly impact v sin i estimates above the typical error of the measurements.

No correlation was found between the v sin i, presence of a disk, and the H-R diagram position. This result indicates that the connection between rotation and the evolution of pre-main sequence stars is both complex and varied.

The RVs of the 10 follow-up targets were measured using a modified CSHELL program, which implements a forward modeling technique. The resulting RVs were analyzed using a

Lomb-Scargle periodogram method to search for periodic signals, potentially indicative of planetary motion. A precision of 68 m s−1 was achieved for the general observations, and

115 m s−1 for the follow-up season that had reduced resolution. The survey is estimated to be sensitive to close-in brown dwarfs, as well as parts of the hot/warm Jupiter regimes.

229 Injection and recovery simulations were implemented using RVSearch (Howard & Fulton

2016) to quantify the sensitivity and set detection limits in each system. This estimate varied widely for each target depending on several factors, such as the number of observations or the level of RV scatter for each individual system. Generally, this survey is sensitive to gas giant planets with mass >6 MJ and brown dwarfs within ∼1 au.

No potential planetary companions were identified in the dataset. Previously, CI Tau

RVs have shown significant periodicity at a 9-day period, as reported by both Johns-Krull et al. (2016), who attributed this signal to a hot Jupiter candidate, and Donati et al. (2020), who argued that this was instead the stellar rotation period. The IGRINS RVs, however, do not confirm this period, and only show a statistically insignificant peak at 9 days in the

IGRINS/LDT data (with a FAP∼6%). Injection/recovery simulations estimated a detection

probability of ∼85-95% for a hot Jupiter companion with the specified parameters. Thus,

there is no evidence in the IGRINS RVs obtained in this survey to confirm this candidate.

Another hot Jupiter candidate, V830 Tau b, could not be confirmed given the survey

sensitivity. However, the rotation period of V830 Tau and the likely rotation period of

DK Tau were recovered at high significance (2.74 days with a FAP of 0.3%, and 8.17 days

with a FAP of 0.4%, respectively). This result confirms that pre-main sequence stars have a

significant level of magnetic activity modulated at the rotation period of the star and that

an infrared survey can still detect activity-induced RV variability. Furthermore, the lack

of a significant periodic signal in the V1075 Tau dataset suggests that the stellar rotation

period cannot always be identified in the infrared data. No significant periodic signals were

found in the K2 target datasets (AA Tau, DM Tau, GI Tau, GM Tau, IQ Tau, LkCa 15), all

of which were identified as possible targets of interest to study star-planet-disk interactions.

This was possibly due, in part, to limited data and resolution since these targets were

primarily observed during the follow-up season.

Injection and recovery simulations using RVSearch placed limits on detections in the

observed systems by estimating detection probabilities of companions with various masses

and separations. The results of these simulations were used to define upper limits on the

230 frequency of substellar companions at a given Mp sin i and semi-major axis. An upper limit

occurrence rate of 25-50% was set for hot Jupiters with masses above 7 MJ, depending on orbital separation. For close-in brown dwarfs, an upper limit on the occurrence rate was

estimated to be between 10-14% for brown dwarfs with masses above 20 MJ at 0.03 au, and

up to 20-50% within 0.3 au. Upper limits for brown dwarfs with masses above 50 MJ and orbital separations out to 0.4 au are between 25-50%.

Observational and theoretical evidence suggests that close-in giant planets and brown

dwarfs may be more prevalent around pre-main sequence stars. Disk migration simulations

indicating a hot Jupiter formation occurrence rate significantly higher than that observed in

main sequence systems (Heller 2019), predictions of brown dwarf migration and destruction

mechanisms (Armitage & Bonnell 2002), and the number of possible companion detections

despite small sample sizes (Quinn et al. 2014; Donati et al. 2017; Yu et al. 2017), all

support this scenario. To test these theories, a comparison of occurrence rates between

main sequence and pre-main sequence samples is necessary. The upper limits estimated

with this survey are consistent with a higher occurrence rate for young close-in brown

dwarfs and gas giant planets, indicating that these predictions could be correct. However,

a scenario where the pre-main sequence and main sequence companion frequencies match is

also possible given that only upper limits could be determined. A larger sample is needed

to confirm whether there is a significant difference in occurrence rates with age. If a larger

sample reveals a young, close-in substellar companion frequency below or consistent with

the main sequence occurrence rate, this could suggest that planet formation or migration

takes longer than the age of the observed systems and that the brown dwarf desert may be

intrinsic to the formation process.

Radial velocities of young stars are challenging to analyze because of the large amount

of RV scatter in the data, primarily caused by magnetic activity. However, this jitter can

also be informative. The average infrared RV scatter of this sample is 550 m s−1, which

is about half the typical optical RV scatter measured in systems that are ∼1-3 Myr old

(∼1 km s−1). This scatter is seemingly related to stellar activity, as it affects both the

231 CTTSs and WTTSs at a similar level. However, the WTTSs are also faster rotators than the CTTSs. Likely both are contributing factors, as stellar rotation and magnetic activity are closely linked. Additional factors related to the disk were also considered. Targets with disk cavities were noted as having lower than average infrared RV jitter (<400 m s−1), and a target identified in the photometry as likely having significant disk accretion also had higher than average RV scatter (∼700 m s−1).

Essentially, young stellar systems are complex environments and create many challenges

for detecting planetary objects. Some of these challenges can be used to learn more about

the systems, such as characterizing the stellar properties and relating them to evolution,

identifying the stellar rotation period from infrared RVs, or quantifying the RV scatter and

searching for correlations with other system properties. Infrared RVs present less jitter

than optical RVs, and with the addition of photometry and an identifiable stellar rotation

period, it is possible to search for young substellar companions in these systems. Detections

of young substellar companions would contribute significantly to formation and evolution

theory related to hot Jupiter migration scenarios or the origin of the brown dwarf desert.

Even in the case of non-detections, injection and recovery simulations offer an opportunity

to place upper limits on the occurrence rates of these objects, which can greatly contribute

to our understanding of planet formation and evolution.

There are several opportunities to expand studies of young planetary objects. Additional

high-resolution infrared spectrographs, such as iSHELL on the NASA Infrared Telescope

Facility and IRD (InfRared Doppler instrument) on the Subaru Telescope, and the infrared

spectropolarimeter, SPIRou on the Canada-France-Hawaii Telescope, are all capable of

observing young potential host stars for RV surveys. These surveys can also be implemented

in conjunction with other techniques. For example, connecting inner planet formation to

outer companions detected from direct imaging, or disk features imaged by ALMA (Atacama

Large Millimeter/submillimeter Array) can provide a system overview of formation and

evolution processes. The space-based TESS mission (Transiting Exoplanet Survey Satellite)

can also be used to determine young candidates for RV follow-up. In addition to companion

232 detection surveys, there is still much progress to be made characterizing RV jitter in infrared observations, and testing combinations of methods to determine the source of RV signals.

Photometrically-derived stellar rotation periods and activity indicators can aid in this effort.

A better characterization of the sources of RV jitter and a fuller picture of these young systems will ultimately enable confirmed detections of substellar companions in short-period orbits.

233 References

Armitage, P. J. & Bonnell, I. A. 2002, MNRAS, 330, L11

Donati, J. F., Bouvier, J., Alencar, S. H., Moutou, C., Malo, L., Takami, M., M´enard,F.,

Dougados, C., Hussain, G. A., & The Matysse Collaboration. 2020, MNRAS, 491, 5660

Donati, J.-F., Yu, L., Moutou, C., Cameron, A. C., Malo, L., Grankin, K., H´ebrard,E.,

Hussain, G. A. J., Vidotto, A. A., Alencar, S. H. P., Haywood, R. D., Bouvier, J., Petit,

P., Takami, M., Herczeg, G. J., Gregory, S. G., Jardine, M. M., & Morin, J. 2017,

MNRAS, 465, 3343

Heller, R. 2019, A&A, 628, A42

Howard, A. W. & Fulton, B. J. 2016, PASP, 128, 114401

Johns-Krull, C. M., McLane, J. N., Prato, L., Crockett, C. J., Jaffe, D. T., Hartigan, P. M.,

Beichman, C. A., Mahmud, N. I., Chen, W., Skiff, B. A., Cauley, P. W., Jones, J. A., &

Mace, G. N. 2016, ApJ, 826, 206

Quinn, S. N., White, R. J., Latham, D. W., Buchhave, L. A., Torres, G., Stefanik, R. P.,

Berlind, P., Bieryla, A., Calkins, M. C., Esquerdo, G. A., F˝ur´esz,G., Geary, J. C., &

Szentgyorgyi, A. H. 2014, ApJ, 787, 27

Yu, L., Donati, J.-F., H´ebrard,E. M., Moutou, C., Malo, L., Grankin, K., Hussain, G.,

Collier Cameron, A., Vidotto, A. A., Baruteau, C., Alencar, S. H. P., Bouvier, J., Petit,

234 P., Takami, M., Herczeg, G., Gregory, S. G., Jardine, M., Morin, J., M´enard,F., &

Matysse Collaboration. 2017, MNRAS

235 Appendix A

Radial Velocities of Pre-Main Sequence Stars and

the RV Standard Star

Table A.1: GJ 281 RVs

JD - 2450000 RV (km s−1) Uncertainty (km s−1)

6984.863 20.219 0.071

6986.015 20.133 0.069

6990.012 20.174 0.067

7334.019 20.235 0.069

7336.042 20.204 0.069

7339.010 20.181 0.069

7742.946 20.263 0.071

7760.922 20.208 0.070

7782.803 20.256 0.070

7799.784 20.121 0.069

8008.022 20.171 0.069

8009.018 20.181 0.070

8010.013 20.204 0.069

8013.021 20.239 0.069

Continued on next page 236 8027.042 20.223 0.069

8028.035 20.224 0.069

8029.037 20.231 0.069

8030.019 20.335 0.070

8031.008 20.274 0.070

8032.004 20.309 0.070

8033.024 20.197 0.070

8033.983 20.147 0.070

8035.023 20.342 0.069

8040.001 20.179 0.069

8046.026 20.288 0.070

8049.038 20.316 0.069

8056.036 20.204 0.069

8056.923 20.185 0.070

8077.947 20.267 0.071

8080.042 20.385 0.070

8081.044 20.179 0.069

8082.049 20.291 0.070

8098.024 20.272 0.070

8101.953 20.214 0.070

8102.976 20.264 0.069

8120.809 20.036 0.069

8124.813 20.163 0.070

8133.798 20.168 0.069

8142.807 20.124 0.069

8143.788 20.242 0.069

Continued on next page

237 8385.020 20.108 0.117

8391.970 20.429 0.117

8412.033 20.378 0.117

8417.049 20.391 0.116

8421.971 20.393 0.117

8422.991 20.032 0.117

8424.033 20.267 0.116

8448.928 20.084 0.116

8451.017 20.398 0.151

8472.808 20.371 0.117

8473.784 20.346 0.116

8502.749 20.216 0.116

8510.788 20.285 0.117

8511.802 20.250 0.116

8512.722 20.329 0.117

8514.690 20.353 0.116

8524.809 20.377 0.116

8526.682 20.329 0.117

8527.775 20.482 0.130

8541.741 20.384 0.119

8547.704 20.164 0.117

8548.700 20.275 0.116

8552.696 20.179 0.116

8559.698 20.271 0.117

238 Table A.2: CI Tau RVs

JD - 2450000 RV (km s−1) Uncertainty (km s−1)

6925.897 16.429 0.110

6940.829 17.225 0.092

6984.776 17.842 0.092

6985.924 17.548 0.091

6986.908 17.316 0.089

6987.851 17.188 0.103

6988.855 17.740 0.085

6989.788 17.300 0.084

6990.814 16.948 0.091

6991.705 16.973 0.081

6992.686 17.380 0.098

6993.745 17.891 0.081

6996.820 17.915 0.079

6997.685 17.798 0.103

7029.576 17.953 0.091

7665.882 16.936 0.089

7666.863 17.323 0.087

7667.894 17.252 0.111

7677.881 17.754 0.075

7678.853 16.684 0.105

7679.888 16.496 0.093

7680.802 17.041 0.112

7703.834 17.448 0.073

7704.848 16.861 0.073

Continued on next page

239 7707.754 17.557 0.072

7707.695 17.635 0.101

7717.916 16.828 0.071

7718.730 16.952 0.071

7730.811 17.445 0.071

7742.650 16.528 0.083

7760.900 17.119 0.090

7782.624 17.254 0.092

7799.618 16.990 0.072

7800.657 17.543 0.072

7800.635 17.459 0.084

8002.943 16.570 0.082

8003.915 16.951 0.071

8007.923 17.008 0.072

8008.961 17.437 0.073

8009.925 17.828 0.073

8012.940 16.194 0.072

8038.882 16.381 0.075

8055.894 16.386 0.108

8056.872 16.253 0.118

8069.770 16.727 0.104

8070.747 17.169 0.294

8073.656 15.828 0.260

8077.705 17.139 0.104

8080.980 17.956 0.106

8097.767 17.143 0.092

Continued on next page

240 8101.877 17.320 0.113

8102.895 16.341 0.115

8131.598 17.190 0.103

8389.886 16.776 0.146

8390.896 16.453 0.142

8391.947 17.173 0.135

8412.045 17.585 0.155

8421.965 17.658 0.136

8423.911 17.557 0.140

8473.602 16.544 0.145

8476.746 16.982 0.128

8502.584 17.079 0.130

8505.660 17.882 0.153

8510.610 16.936 0.131

8512.689 15.515 0.386

8514.592 17.398 0.131

8516.617 16.374 0.129

8524.606 17.476 0.130

8526.618 16.662 0.128

8527.616 16.719 0.130

8547.643 17.265 0.133

8552.614 16.784 0.141

8557.660 17.315 0.143

8559.613 16.855 0.133

241 Table A.3: V830 Tau RVs

JD - 2450000 RV (km s−1) Uncertainty (km s−1)

6989.845 16.912 0.127

7288.900 17.193 0.161

7289.870 17.729 0.130

7291.926 16.982 0.145

7296.860 17.340 0.128

7298.003 17.268 0.120

7331.949 16.476 0.128

7332.827 17.169 0.143

7333.705 17.245 0.131

7336.776 17.284 0.136

7338.981 16.930 0.152

7665.917 16.293 0.126

7666.854 16.937 0.119

7667.927 16.837 0.132

7677.915 17.173 0.126

7678.829 17.865 0.136

7679.840 17.112 0.077

7680.879 16.909 0.074

7703.855 17.117 0.082

7707.705 16.766 0.075

7717.845 16.893 0.080

7718.776 16.906 0.079

7730.772 16.753 0.073

7760.755 16.891 0.075

Continued on next page

242 7782.612 16.797 0.119

7800.694 16.685 0.120

8002.992 16.852 0.076

8003.981 16.584 0.074

8007.878 17.256 0.076

8008.901 17.086 0.107

8009.879 17.536 0.076

8012.881 17.484 0.073

8026.940 17.231 0.073

8027.909 17.146 0.075

8029.888 17.207 0.074

8030.911 17.133 0.073

8031.902 17.473 0.076

8032.924 16.665 0.075

8033.882 16.332 0.123

8038.918 17.107 0.074

8045.966 18.017 0.074

8055.863 16.929 0.077

8056.849 17.652 0.077

8069.766 16.536 0.143

8070.725 18.155 0.215

8072.690 16.564 0.105

8077.699 16.703 0.140

8078.864 17.283 0.105

8078.934 17.216 0.141

8080.924 16.455 0.112

Continued on next page

243 8097.772 16.637 0.107

8101.895 17.346 0.179

8102.909 16.258 0.155

8131.592 17.163 0.138

8389.866 17.105 0.169

8391.847 17.687 0.179

8411.843 16.860 0.152

8421.956 17.538 0.157

8422.964 16.617 0.138

8447.871 16.553 0.134

8450.931 16.250 0.161

8473.569 16.552 0.155

8476.728 17.294 0.157

8483.801 16.062 0.154

8502.575 16.263 0.149

8503.716 16.702 0.166

8506.576 16.601 0.156

8507.584 15.790 0.171

8508.579 16.018 0.153

8509.573 16.988 0.144

8510.620 16.182 0.169

8511.592 16.305 0.154

8512.593 16.638 0.158

8513.592 16.129 0.198

8514.586 16.561 0.156

8516.599 16.194 0.158

Continued on next page

244 8524.596 16.096 0.150

8525.617 16.702 0.136

8526.610 16.036 0.149

8527.603 16.343 0.156

8547.632 16.654 0.146

8548.607 16.156 0.160

8552.606 16.437 0.159

8557.622 15.960 0.173

8559.607 16.168 0.147

245 Table A.4: DK Tau RVs

JD - 2450000 RV (km s−1) Uncertainty (km s−1)

6985.907 17.017 0.143

7707.935 14.957 0.101

7742.880 16.479 0.097

8025.905 15.016 0.098

8027.948 14.925 0.113

8028.882 17.139 0.109

8030.871 13.109 0.226

8031.880 15.550 0.233

8032.898 14.807 0.166

8034.857 16.251 0.126

8035.939 15.794 0.129

8038.845 13.031 0.221

8046.004 16.865 0.115

8048.883 14.653 0.134

8055.819 15.358 0.131

8056.832 15.561 0.198

8069.760 16.344 0.213

8070.713 17.799 0.348

8072.675 15.469 0.209

8073.646 14.492 0.129

8074.753 14.450 0.110

8077.687 16.165 0.125

8078.840 16.887 0.141

8079.810 16.247 0.119

Continued on next page

246 8082.041 16.476 0.096

8097.762 13.718 0.153

8101.885 16.541 0.151

8102.901 16.907 0.115

8120.609 15.225 0.130

8124.572 15.798 0.107

8131.585 15.303 0.099

8133.586 15.719 0.089

8389.877 16.053 0.211

8390.907 14.539 0.159

8391.952 15.572 0.152

8421.945 17.294 0.222

8422.924 12.495 0.265

8423.967 13.566 0.217

8442.662 15.704 0.244

8472.796 14.637 0.150

8476.710 15.829 0.149

8502.591 16.219 0.139

8510.602 16.415 0.134

8514.598 14.651 0.140

8516.684 15.085 0.129

8524.614 15.702 0.143

8525.729 15.530 0.126

8526.624 16.391 0.164

8527.631 16.129 0.152

8547.610 15.687 0.134

Continued on next page

247 8548.621 15.598 0.130

8552.622 17.500 0.161

8558.675 15.929 0.154

8559.620 16.273 0.127

248 Table A.5: V1075 Tau RVs

JD - 2450000 RV (km s−1) Uncertainty (km s−1)

6991.924 17.767 0.198

7665.894 18.505 0.145

7666.827 18.225 0.135

7667.907 18.433 0.175

7677.907 18.272 0.149

7678.900 18.839 0.231

7679.907 18.517 0.152

7680.837 18.220 0.164

7703.883 18.304 0.144

7704.930 18.461 0.150

7707.738 18.310 0.117

7717.764 17.880 0.143

7718.699 18.060 0.154

7730.709 17.899 0.132

7742.859 18.367 0.352

7759.883 18.108 0.140

7782.658 18.159 0.152

7799.770 18.121 0.188

7800.611 17.785 0.125

8002.910 18.981 0.355

8003.888 18.948 0.136

8004.981 18.276 0.152

8007.957 18.889 0.138

8008.986 17.971 0.194

Continued on next page

249 8009.958 17.961 0.134

8012.969 18.653 0.145

8025.884 18.358 0.144

8026.899 17.990 0.127

8027.872 18.389 0.123

8028.868 18.764 0.157

8029.944 18.367 0.176

8030.953 18.028 0.167

8031.932 18.461 0.196

8032.951 17.937 0.159

8033.921 17.457 0.203

8038.863 17.924 0.153

8039.858 18.429 0.117

8046.018 18.924 0.169

8048.891 18.210 0.170

8055.908 18.369 0.155

8056.891 18.378 0.153

8069.781 18.299 0.145

8072.700 18.088 0.129

8073.714 18.563 0.120

8074.762 18.437 0.142

8077.720 17.691 0.140

8078.851 18.323 0.118

8079.826 17.949 0.116

8080.900 18.349 0.167

8082.013 18.279 0.148

Continued on next page

250 8097.784 17.844 0.126

8101.853 17.427 0.195

8102.880 18.400 0.152

8131.615 17.983 0.144

8384.925 18.166 0.195

8389.895 18.334 0.186

8390.816 19.522 0.182

8391.817 17.966 0.234

8411.814 18.386 0.164

8417.021 18.541 0.156

8421.002 18.532 0.176

8421.823 18.192 0.182

8422.866 18.310 0.204

8423.804 18.114 0.193

8449.827 20.476 0.161

8450.779 18.097 0.157

8451.699 18.487 0.151

8473.609 15.928 0.191

8476.869 17.960 0.153

8502.556 18.018 0.181

8503.727 17.225 0.188

8504.581 17.854 0.385

8505.627 17.581 0.182

8506.597 17.249 0.175

8507.593 17.634 0.193

8508.589 17.192 0.174

Continued on next page

251 8509.588 18.047 0.197

8510.592 17.435 0.182

8511.583 17.891 0.169

8512.563 18.246 0.172

8513.624 17.626 0.188

8514.576 17.820 0.188

8516.581 17.597 0.177

8524.574 18.387 0.188

8525.581 17.698 0.140

8526.593 18.017 0.162

8527.580 17.534 0.155

8547.583 17.359 0.155

8548.589 17.728 0.160

8552.592 17.501 0.180

8557.598 17.918 0.150

8558.623 17.213 0.268

8559.599 17.275 0.181

252 Table A.6: AA Tau RVs

JD - 2450000 RV (km s−1) Uncertainty (km s−1)

6990.637 16.971 0.111

7028.645 17.033 0.084

7386.794 18.126 0.125

7387.852 18.543 0.124

7707.980 17.024 0.094

7711.865 17.726 0.109

7713.054 18.172 0.156

7713.063 17.784 0.150

7715.891 18.772 0.127

7716.824 17.875 0.099

7800.730 17.298 0.081

8056.943 17.870 0.088

8081.037 18.539 0.106

8384.944 17.705 0.142

8389.901 18.562 0.169

8391.857 17.475 0.131

8411.894 17.628 0.125

8421.846 17.036 0.128

8423.023 17.088 0.126

8465.826 17.639 0.129

8473.580 17.399 0.125

8476.717 18.098 0.121

8502.633 18.724 0.125

8510.666 19.481 0.159

Continued on next page

253 8512.636 17.673 0.211

8514.618 17.154 0.124

8524.679 17.961 0.128

8525.744 18.012 0.127

8526.654 18.447 0.130

8527.692 18.236 0.172

8547.687 17.403 0.125

8548.672 18.139 0.123

8552.661 17.824 0.138

8557.687 17.390 0.128

254 Table A.7: DM Tau RVs

JD - 2450000 RV (km s−1) Uncertainty (km s−1)

7419.573 18.457 0.078

7718.815 18.385 0.074

8056.906 18.251 0.073

8072.758 18.291 0.072

8120.674 18.289 0.073

8384.958 18.255 0.121

8390.837 18.418 0.122

8391.871 18.334 0.122

8411.912 18.188 0.118

8421.857 18.277 0.121

8423.834 18.324 0.122

8502.683 18.512 0.120

8510.703 18.414 0.124

8514.652 18.463 0.119

8524.748 18.574 0.120

255 Table A.8: GI Tau RVs

JD - 2450000 RV (km s−1) Uncertainty (km s−1)

6986.979 17.923 0.090

6997.697 17.303 0.096

7708.864 17.269 0.081

8056.021 17.165 0.092

8073.788 17.073 0.103

8384.971 16.958 0.129

8389.908 17.472 0.149

8391.886 17.176 0.124

8411.928 15.963 0.151

8421.864 17.458 0.138

8423.852 17.326 0.142

8450.678 18.348 0.130

8473.597 17.388 0.137

8476.723 17.796 0.137

8502.672 17.254 0.134

8510.677 16.447 0.143

8512.707 17.277 0.145

8514.643 17.571 0.146

8516.690 17.274 0.155

8524.717 16.909 0.135

8547.668 17.691 0.129

8559.692 16.783 0.149

256 Table A.9: GM Tau RVs

JD - 2450000 RV (km s−1) Uncertainty (km s−1)

7667.860 16.181 0.155

7730.919 16.341 0.213

8035.005 15.710 0.119

8097.833 16.186 0.110

8114.774 16.766 0.137

8384.993 16.874 0.217

8390.856 16.238 0.283

8391.897 16.377 0.211

8411.954 16.989 0.291

8421.882 15.462 0.183

8423.936 16.773 0.255

8473.673 15.720 0.257

8476.843 16.385 0.158

8502.718 16.043 0.176

8510.742 15.695 0.263

8514.670 17.249 0.156

8524.788 15.162 0.163

8559.663 14.985 0.184

257 Table A.10: IQ Tau RVs

JD - 2450000 RV (km s−1) Uncertainty (km s−1)

6940.892 15.827 0.118

6996.795 16.349 0.092

7379.745 16.905 0.082

7707.958 15.333 0.101

8056.828 16.047 0.134

8077.751 14.046 0.139

8133.809 15.863 0.103

8389.822 16.137 0.150

8390.886 15.368 0.151

8391.932 14.667 0.138

8412.010 16.032 0.134

8421.932 15.625 0.133

8423.878 14.212 0.133

8450.747 15.884 0.129

8473.592 15.828 0.134

8476.706 15.464 0.139

8502.667 15.322 0.136

8510.657 15.844 0.134

8514.638 14.318 0.137

8516.695 15.696 0.129

8524.707 15.873 0.128

8526.671 15.327 0.137

8527.746 15.461 0.196

8547.601 15.033 0.131

Continued on next page

258 8559.683 14.200 0.166

259 Table A.11: LkCa 15 RVs

JD - 2450000 RV (km s−1) Uncertainty (km s−1)

6990.668 18.099 0.131

6990.693 17.709 0.137

7364.605 17.846 0.107

7671.933 17.777 0.122

7672.883 17.875 0.131

7675.907 18.239 0.096

7708.969 17.940 0.096

8056.015 17.898 0.100

8079.864 18.499 0.096

8124.676 18.074 0.109

8389.846 17.631 0.162

8390.892 18.003 0.158

8391.938 17.490 0.139

8412.020 18.064 0.145

8421.939 17.460 0.135

8423.897 18.763 0.142

8473.632 18.065 0.131

8476.737 17.866 0.140

8502.660 17.157 0.133

8510.687 18.009 0.140

8512.743 17.075 0.352

8514.632 17.923 0.150

8516.702 17.548 0.136

8524.697 17.996 0.146

Continued on next page

260 8526.665 17.742 0.156

8527.715 18.318 0.202

8547.658 17.591 0.134

8548.689 17.761 0.176

8552.685 17.567 0.140

8559.706 18.483 0.141

261