Understanding Hot Jupiters with the Dedicated Monitor of Exotransits and Transients (DEMONEXT)

Understanding Hot Jupiters with the DEdicated MONitor of EXotransits and Transients (DEMONEXT)

DISSERTATION

Presented in Partial Fulﬁllment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Steven Villanueva Jr.

Graduate Program in Astronomy

The Ohio State University 2018

Dissertation Committee: Dr. B. Scott Gaudi, Advisor Dr. Richard Pogge, Co-Advisor Dr. Laura Lopez Copyright by

Steven Villanueva Jr.

2018 Abstract

We present results for the misaligned Hot Jupiter, XO-4b, taken with the decommissioned DEdicated MONitor of EXotransits (DEMONEX) telescope, as a proof-of-concept for the upcoming DEdicated MONitor of EXotransits and

Transients (DEMONEXT) telescope. We combine these data with archival light curves and archival radial velocity measurements to derive the XO-4 host star mass

+0.030 +0.042 M∗ = 1.293−0.029 M and radius R∗ = 1.554−0.030 R as well as the XO-4b planet

+0.10 +0.040 mass MP = 1.615−0.099 MJ and radius RP = 1.317−0.029 RJ and a reﬁned ephemeris of P = 4.1250687 ± 0.0000024 days and T0 = 2454758.18978 ± 0.00024 BJDTDB. We include archival Rossiter-McLaughlin measurements of XO-4 to infer the stellar spin-

+8.8 planetary orbit alignment λ = −40.0−7.5 degrees. In preparation for DEMONEXT observations, we test the effects of including various detrend parameters, theoretical and empirical mass-radius relations, and Rossiter-McLaughlin models. We infer that detrending against CCD position and time or airmass can improve data quality, but can have significant effects on the inferred values of many parameters — most significantly RP /R∗ and the observed central transit times TC . In the case of

RP /R∗ we ﬁnd that the systematic uncertainty due to detrending can be three times that of the quoted statistical uncertainties. The choice of mass-radius relation

ii has little eﬀect on our inferred values of the system parameters, but the choice of

Rossiter-McLaughlin models can have signiﬁcant eﬀects of the inferred values of v sin I∗ and the stellar spin-planet orbit angle λ.

We report on the design and ﬁrst two years of operations of the newly commissioned DEMONEXT telescope. DEMONEXT is a 20 inch (0.5-m) robotic telescope using a PlaneWave CDK20 telescope on a Mathis instruments MI-750/1000 fork mount. DEMONEXT is equipped with a 2048 × 2048 pixel Finger Lakes

Instruments (FLI) detector, a 10-position ﬁlter wheel with an electronic focuser and

B, V , R, I, g0, r0, i0, z0, and clear ﬁlters. DEMONEXT operates in a continuous observing mode and achieves 2–4 mmag raw, unbinned, precision on bright V < 13 targets with 20–120 second exposures, and 1 mmag precision achieved by binning on 5–6 minute timescales. DEMONEXT maintains sub-pixel (< 0.5 pixels) target position stability on the CCD over 8 hours in good observing conditions, with degraded performance in poor weather (< 1 pixel). DEMONEXT achieves 1–10% photometry on single-epoch targets with V < 17 in 5 minute exposures, with detection thresholds of V ≈ 21. The DEMONEXT automated software has produced over 300 planetary candidate transit light curves for the KELT collaboration, and over 100 supernovae and transient light curves for the ASAS-SN supernovae group in the ﬁrst two years of operations. DEMONEXT has also observed for a number of ancillary science projects including Galactic microlensing, active galactic nuclei, stellar variability, and stellar rotation.

iii Finally, we present a semi-analytic estimate of the expected yield of single- transit planets from TESS as possible future targets for DEMONEXT. We use the TESS Candidate Target List 6 (CTL-6) as an input catalog of over 4 million sources. We predict that from the 200,000 stars selected to be observed with the high-cadence postage stamps with the highest CTL-6 priority, there will be

241 single-transit events caused by planets detectable at a signal-to-noise ratio of

SNR≥ 7.3. We ﬁnd a lower limit of an additional 977 events caused by single-transit planets in the full frame images (FFI); this is a lower limit because the CTL-6 is incomplete below a TESS magnitude of T > 12. Of the single-transit events from the postage stamps and FFIs, 1091/1218 will have transit depths deeper than

0.1%, and will thus be amenable for photometric follow-up from the ground from telescopes like DEMONEXT. 1195/1218 will have radial velocity signals greater than 1 m/s, measurable from current ground-based telescopes. We estimate that the periods of 146 single transits will be constrained to better than 10% using the TESS photometry assuming circular orbits. We ﬁnd that the number of planets detected by TESS in the postage stamps with periods P > 25 days will be doubled by including single-transiting planets, while the number of planets with P > 250 days will be increased by an order of magnitude. We predict 79 habitable zone planets from single-transits, with 18 orbiting FGK stars.

iv Dedication

To all of those more talented, more deserving, and harder working than I,

who’ve never been aﬀorded the opportunities I’ve had.

v Acknowledgments

It is very unfortunate that only one person may claim credit for this dissertation, as there are many more people responsible for this work than I could possibly name or remember. To those I unintentionally left off, remember that I took me six years to write the bulk of this, and I only had a few months to finish it off and write the introductions.

First and foremost, this dissertation is for my family. I have to acknowledge their support and sacrifices that go back to before I was born, much less a graduate student. For my mother, who’s sacrifices throughout my life put me in a position to have a fighting chance. To say that she is more responsible for this work than any one else is an understatement. For my sisters, Steffani and Staci, if I do anything in life it is so that the path is easier for any who may follow me, especially for the two of you. The same goes for my nieces, Laila and Lana. I hope you two outshine us all. My dad, who has been a source of constant support and my biggest fan.

To my Nana and Papa Vallejo, who bought me my ﬁrst computer and took me to the Johnston Space Center. I still own that ﬂight suit. To Grandma and Grandpa

Villanueva, all the months and years away from Texas are erased the moment you

vi serve me a plate of your rice. To all my aunts and uncles, thank you for your love and support. For my seemingly endless list of cousins, thank you for making opening

Christmas presents take hours. A special thanks to Mark and Teresa, who made their home in Chicago my home away from home during grad school.

A special thank you to all the friends who might as well be family. Gabe

Mendoza, who would have thought a road trip would have sparked so much. Patrick

Williams, let’s play some more FIFA. Calen B. Henderson, I don’t really have to words to thank you for being who you are. I do have one word, but you’ll have to get the .tex ﬁle.penis

To the graduate students and postdocs at OSU for making grad school what it is, especially my classmates Dan Stevens, Jamie Tayar, and Michael Fausnaugh.

I’ll see you at Hounddog’s at 7? Can’t forget my Adventurers, Andres, Carl, Jamie,

Mathias, Tyler, and Sam. To everyone who ever sportsballed with me: Sarah, Matt,

Garrett, Bret, Sun, Thomas, Kate, Jill, Carl, Jamie, and the rest of the roster. For those who arrived late and I didn’t get to spend enough time with, especially Sam and Romy. J-Tan, may your MMR only go up from here.

I’d like to thank all of the faculty, especially Todd Thompson and Laura Lopez.

Todd, you were there for me when no one else was, and looked after me as if I was one of your own. I don’t think I can thank you enough. Thank you Laura for

vii just being you, for you constant support of SACNAS, Diversity Journal Club, and connecting me with some amazing people. Please be in charge of colloquium forever.

A huge thanks to everyone in the OSU SACNAS Chapter for giving me a second home. Especially Marcella Hernandez, who taught me all the things that aren’t taught in grad school but should be. Shaun Hampton, you are my SACNAS brother for life.

To all of the astronomers of color that have made an impact of me these past few years, thank you. Especially those that have for paved the way. Jorge Moreno and Jes´usPando are my heroes.

Science doesn’t get done without someone paying the bills. Thank you to the the David G. Price Fellowship for Astronomical Instrumentation and the National

Science Foundation Graduate Research Fellowship under Grant No. DGE-1343012.

I would also like to thank all the money my co-authors brought in including:

Scott’s funding and a signiﬁcant fraction of the DEMONEXT hardware costs were supported by National Science Foundation CAREER Grant AST-1056524;

Diana Dragomir was supported by NASA through the Hubble Fellowship grant

HST-HF2-51372.001-A awarded by the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., for NASA, under contract NAS5-26555; Keivan Stassun and DEMONEXT is also supported from the Vanderbilt Initiative in Data-intensive Astrophysics (VIDA).

viii To all of my co-authors not yet mentioned, especially Jason Eastman and Mark

& Pat Trueblood. Thanks to PlaneWave Instruments all thier help, especially Rick

Hedrick and Kevin Ivarson. I would like to thank Jerry Mason, Tom OBrien, Dan

Pappalardo, Jon Shover, Dave Steinbrecher, and the rest of the staﬀ of the OSU

Imaging Sciences Laboratory. Of course nothing gets done without David Will.

Thank you to everyone that provided help with targeting, scheduling, and shared their preliminary results including members of the KELT collaboration:

Jonathan Labadie-Bartz, Mike Lund, Josh Pepper, Joey Rodriquez, and Dan

Stevens; the ASAS-SN team: Subhash Bose, Jon Brown, Ping Chen, Subo Dong, and Tom Holoien; the Open Clusters team: Ryan Oelkers and Garrett Sommers; and the microlensing team: Andy Gould, Money Penny, Jen Yee, and Wei Zhu.

A big thanks to team MIT and team TESS for helping with the transition, especially Diana Dragomir, Chelsea Huang, Ian Crossﬁeld, Tom Barclay, and Luke

Bouma.

None of this work would have gotten done with out the help and support of my advisors. Rick, Jen and Darren had nothing but great things to say about you, and you’ve lived up to all the hype and then some. Scott, I’m going to quote your own dissertation acknowledgements here: ”I may have been able to choose an easier adviser; I could not have chosen a better one.” Thank you for tolerating, enduring,

ix and supporting me through my graduate career, especially when it came to all of the non-research things.

Last and certainly not least, to mi amor. Jennifer you have been there for me more than anyone else during these past four years. Your love and support kept me going, even if I thought otherwise. I love you to the furthest exoplanet and back.

That goes for Lunita too.

x Vita

August 2, 1984...... Born – Dallas, TX, USA

2003 – 2007 ...... Aerospace Ground Equipment Mechanic,

United States Air Force, RAF Lakenheath

2012 ...... B.S. Physics, Texas A&M University

2012 – 2014 ...... Graduate Teaching and Research Associate,

The Ohio State University

2014 – 2017 ...... National Science Foundation Graduate

Research Fellow, The Ohio State University

2016 ...... M.S. Astronomy, The Ohio State University

2017 – 2018 ...... David G. Price Fellow in Astronomical

Instrumentation, The Ohio State University

Publications

Research Publications

1. S. Villanueva Jr., D. Dragomir, and B. S. Gaudi, ”An Estimate of the Yield of Single-Transit Planetary Events from the Transiting Exoplanet Survey Satellite”, ArXiv e-prints, arXiv:1805.00956, (2018)

2. J. Labadie-Bartz, J. E. Rodriguez, K. G. Stassun, D. R. Ciardi, M. C.

xi Johnson, B. S. Gaudi, K. M. Penev, A. Bieryla, D. W. Latham, J. Pepper, K. A. Collins, P. Evans, H. M. Relles, R. J. Siverd, J. Bento, X. Yao, C. Stockdale, T.-G. Tan, G. Zhou, K. D. Colon, J. D. Eastman, M. D. Albrow, A. Malpas, D. Bayliss, T. G. Beatty, V. Bozza, D. H. Cohen, I. A. Curtis, D. L. DePoy, D. Feliz, B. J. Fulton, J. Gregorio, D. James, H. Jang-Condell, E. L. Jensen, J. A. Johnson, S. A. Johnson, M. D. Joner, J. Kielkopf, R. B. Kuhn, M. B. Lund, M. Manner, J. L. Marshall, N. McCrady, K. K. McLeod, T. E. Oberst, M. T. Penny, R. Pogge, P. A. Reed, D. H. Sliski, D. C. Stephens, D. J. Stevens, M. Trueblood, P. Trueblood, S. Villanueva Jr., R. A. Wittenmyer, J. T. Wright, R. Zambelli, P. Berlind, M. L. Calkins, and G. A. Esquerdo, ”KELT-22Ab: A Massive Hot Jupiter Transiting a Near Solar Twin”, ArXiv e-prints, arXiv:1803.07559, (2018)

3. M. M. Fausnaugh, D. A. Starkey, K. Horne, C. S. Kochanek, B. M. Pe- terson, M. C. Bentz, K. D. Denney, C. J. Grier, D. Grupe, R. W. Pogge, G. De Rosa, S. M. Adams, A. J. Barth, T. G. Beatty, A. Bhattacharjee, G. A. Borman, T. A. Boroson, M. C. Bottorﬀ, J. E. Brown, J. S. Brown, M. S. Brotherton, C. T. Coker, S. M. Crawford, K. V. Croxall, S. Eftekharzadeh, M. Eracleous, M. D. Joner, C. B. Henderson, T. W.-S. Holoien, T. Hutchison, S. Kaspi, S. Kim, A. L. King, M. Li, C. Lochhaas, Z. Ma, F. MacInnis, E. R. Manne-Nicholas, M. Mason, C. Montuori, A. Mosquera, D. Mudd, R. Musso, S. V. Nazarov, M. L. Nguyen, D. N. Okhmat, C. A. Onken, B. Ou-Yang, A. Pancoast, L. Pei, M. T. Penny, R. Poleski, S. Rafter, E. Romero-Colmenero, J. Runnoe, D. J. Sand, J. S. Schimoia, S. G. Sergeev, B. J. Shappee, G. V. Simonian, G. Somers, M. Spencer, D. J. Stevens, J. Tayar, T. Treu, S. Valenti, J. Van Saders, S. Villanueva Jr., C. Villforth, Y. Weiss, H. Winkler, and W. Zhu, ”Continuum Reverberation Mapping of the Ac- cretion Disks in Two Seyfert 1 Galaxies”, The Astrophysical Journal, 854, 107, (2018)

4. M. C. Johnson, J. E. Rodriguez, G. Zhou, E. J. Gonzales, P. A. Cargile, J. R. Crepp, K. Penev, K. G. Stassun, B. S. Gaudi, K. D. Col´on,D. J. Stevens, K. G. Strassmeier, I. Ilyin, K. A. Collins, J. F. Kielkopf, T. E. Oberst, L. Maritch, P. A. Reed, J. Gregorio, V. Bozza, S. Calchi Novati, G. D’Ago, G. Scarpetta, R. Zambelli, D. W. Latham, A. Bieryla, W. D. Cochran, M. Endl, J. Tayar, A. Serenelli, V. Silva Aguirre, S. P. Clarke, M. Martinez, M. Spencer, J. Trump, M. D. Joner, A. G. Bugg, E. G. Hintz, D. C. Stephens, A. Arredondo, A. Benzaid, S. Yazdi, K. K. McLeod, E. L. N. Jensen, D. A. Hancock, R. L. Sorber, D. H. Kasper, H. Jang-Condell, T. G. Beatty, T. Carroll, J. Eastman, D. James, R. B. Kuhn, J. Labadie-Bartz, M. B. Lund, M. Mallonn, J. Pepper, R. J. Siverd, X. Yao, D. H. Cohen, I. A. Curtis, D. L. DePoy, B. J. Fulton, M. T. Penny, H. Relles, C. Stockdale, T.-G. Tan, and S. Villanueva Jr., ”KELT-21b: A Hot Jupiter Transiting the Rapidly Rotating Metal-poor Late-A Primary of a Likely Hierarchical Triple System”, The Astronomical Journal, 155, 100, (2018)

xii 5. S. Villanueva Jr., B. S. Gaudi, R. W. Pogge, J. D. Eastman, K. G. Stas- sun, M. Trueblood, and P. Trueblood, ”DEdicated MONitor of EXotransits and Transients (DEMONEXT): System Overview and Year One Results from a Low-cost Robotic Telescope for Followup of Exoplanetary Transits and Transients”, Publications of the Astronomical Society of the Paciﬁc, 130, 015001, (2018)

6. J. S. Brown, C. S. Kochanek, T. W.-S. Holoien, K. Z. Stanek, K. Auchettl, B. J. Shappee, J. L. Prieto, N. Morrell, E. Falco, J. Strader, L. Chomiuk, R. Post, S. Villanueva Jr., S. Mathur, S. Dong, P. Chen, and S. Bose, ”The ultraviolet spectroscopic evolution of the low-luminosity tidal disruption event iPTF16fnl”, Monthly Notices of the Royal Astronomical Society, 473, 1130, (2018)

7. S. Bose, S. Dong, A. Pastorello, A. V. Filippenko, C. S. Kochanek, J. Mauerhan, C. Romero-Ca nizales, T. G. Brink, P. Chen, J. L. Prieto, R. Post, C. Ashall, D. Grupe, L. Tomasella, S. Benetti, B. J. Shappee, K. Z. Stanek, Z. Cai, E. Falco, P. Lundqvist, S. Mattila, R. Mutel, P. Ochner, D. Pooley, M. D. Stritzinger, S. Villanueva Jr., W. Zheng, R. J. Beswick, P. J. Brown, E. Cappellaro, S. Davis, M. Fraser, T. de Jaeger, N. Elias-Rosa, C. Gall, B. S. Gaudi, G. J. Herczeg, J. Hestenes, T. W.-S. Holoien, G. Hosseinzadeh, E. Y. Hsiao, S. Hu, S. Jaejin, B. Jeﬀers, R. A. Koﬀ, S. Kumar, A. Kurtenkov, M. W. Lau, S. Prentice, T. Reynolds, R. J. Rudy, M. Shahbandeh, A. Somero, K. G. Stassun, T. A. Thompson, S. Valenti, J.-H. Woo, and S. Yunus, ”Gaia17biu/SN 2017egm in NGC 3191: The Closest Hydrogen-poor Superluminous Supernova to Date Is in a ’Normal,’ Massive, Metal-rich Spiral Galaxy”, The Astrophysical Journal, 853, 57, (2018)

8. R. J. Siverd, K. A. Collins, G. Zhou, S. N. Quinn, B. S. Gaudi, K. G. Stassun, M. C. Johnson, A. Bieryla, D. W. Latham, D. R. Ciardi, J. E. Rodriguez, K. Penev, M. Pinsonneault, J. Pepper, J. D. Eastman, H. Relles, J. F. Kielkopf, J. Gregorio, T. E. Oberst, G. F. Aldi, G. A. Esquerdo, M. L. Calkins, P. Berlind, C. D. Dressing, R. Patel, D. J. Stevens, T. G. Beatty, M. B. Lund, J. Labadie-Bartz, R. B. Kuhn, K. D. Col´on,D. James, X. Yao, J. A. Johnson, J. T. Wright, N. McCrady, R. A. Wittenmyer, S. A. Johnson, D. H. Sliski, E. L. N. Jensen, D. H. Cohen, K. K. McLeod, M. T. Penny, M. D. Joner, D. C. Stephens, S. Villanueva Jr., R. Zambelli, C. Stockdale, P. Evans, T.-G. Tan, I. A. Curtis, P. A. Reed, M. Trueblood, and P. Trueblood, ”KELT-19Ab: A P ∼ 4.6-day Hot Jupiter Transiting a Likely Am Star with a Distant Stellar Companion”, The Astronomical Journal, 155, 35, (2018)

9. K. Krisciunas, C. Contreras, C. R. Burns, M. M. Phillips, M. Hamuy, M. D. Stritzinger, J. Anais, L. Boldt, L. Busta, A. Campillay, S. Castellón,G. Folatelli, W. L. Freedman, C. González, E. Y. Hsiao, W. Krzeminski, N. Morrell, S. E. Persson, M. Roth, F. Salgado, J. Serón,N. B. Suntzeff, S. Torres, A. V. Filippenko, W. Li, B. F. Madore, D. L. Depoy, J. L. Marshall, J.-P. Rheault, and S. Villanueva,

xiii ”Erratum: ’The Carnegie Supernova Project. I. Third Photometry Data Release of Low-redshift Type Ia Supernovae and Other White Dwarf Explosions’”, The Astronomical Journal, 154, 278, (2017)

10. L. Y. Temple, C. Hellier, M. D. Albrow, D. R. Anderson, D. Bayliss, T. G. Beatty, A. Bieryla, D. J. A. Brown, P. A. Cargile, A. Collier Cameron, K. A. Collins, K. D. Col´on,I. A. Curtis, G. D’Ago, L. Delrez, J. Eastman, B. S. Gaudi, M. Gillon, J. Gregorio, D. James, E. Jehin, M. D. Joner, J. F. Kielkopf, R. B. Kuhn, J. Labadie-Bartz, D. W. Latham, M. Lendl, M. B. Lund, A. L. Malpas, P. F. L. Maxted, G. Myers, T. E. Oberst, F. Pepe, J. Pepper, D. Pollacco, D. Queloz, J. E. Rodriguez, D. S´egransan,R. J. Siverd, B. Smalley, K. G. Stassun, D. J. Stevens, C. Stockdale, T. G. Tan, A. H. M. J. Triaud, S. Udry, S. Villanueva, R. G. West, and G. Zhou, ”WASP-167b/KELT-13b: joint discovery of a hot Jupiter transiting a rapidly rotating F1V star”, Monthly Notices of the Royal Astronomical Society, 471, 2743, (2017)

11. K. Krisciunas, C. Contreras, C. R. Burns, M. M. Phillips, M. D. Stritzinger, N. Morrell, M. Hamuy, J. Anais, L. Boldt, L. Busta, A. Campillay, S. Castellón,G. Folatelli, W. L. Freedman, C. González,E. Y. Hsiao, W. Krzeminski, S. E. Persson, M. Roth, F. Salgado, J. Serón,N. B. Suntzeff, S. Torres, A. V. Filippenko, W. Li, B. F. Madore, D. L. DePoy, J. L. Marshall, J.-P. Rheault, and S. Villanueva, ”The Carnegie Supernova Project. I. Third Photometry Data Release of Low-redshift Type Ia Supernovae and Other White Dwarf Explosions”, The Astronomical Journal, 154, 211, (2017)

12. M. B. Lund, J. E. Rodriguez, G. Zhou, B. S. Gaudi, K. G. Stassun, M. C. Johnson, A. Bieryla, R. J. Oelkers, D. J. Stevens, K. A. Collins, K. Penev, S. N. Quinn, D. W. Latham, S. Villanueva Jr., J. D. Eastman, J. F. Kielkopf, T. E. Oberst, E. L. N. Jensen, D. H. Cohen, M. D. Joner, D. C. Stephens, H. Relles, G. Corﬁni, J. Gregorio, R. Zambelli, G. A. Esquerdo, M. L. Calkins, P. Berlind, D. R. Ciardi, C. Dressing, R. Patel, P. Gagnon, E. Gonzales, T. G. Beatty, R. J. Siverd, J. Labadie-Bartz, R. B. Kuhn, K. D. Col´on,D. James, J. Pepper, B. J. Fulton, K. K. McLeod, C. Stockdale, S. Calchi Novati, D. L. DePoy, A. Gould, J. L. Marshall, M. Trueblood, P. Trueblood, J. A. Johnson, J. Wright, N. McCrady, R. A. Wittenmyer, S. A. Johnson, A. Sergi, M. Wilson, and D. H. Sliski, ”KELT-20b: A Giant Planet with a Period of P ∼ 3.5 days Transit- ing the V ∼ 7.6 Early A Star HD 185603”, The Astronomical Journal, 154, 194, (2017)

13. S. Mathur, A. Gupta, K. Page, R. W. Pogge, Y. Krongold, M. R. Goad, S. M. Adams, M. D. Anderson, P. Ar´evalo, A. J. Barth, C. Bazhaw, T. G. Beatty, M. C. Bentz, A. Bigley, S. Bisogni, G. A. Borman, T. A. Boroson, M. C. Bottorﬀ, W. N. Brandt, A. A. Breeveld, J. E. Brown, J. S. Brown, E. M. Cackett, G. Canalizo,

xiv M. T. Carini, K. I. Clubb, J. M. Comerford, C. T. Coker, E. M. Corsini, D. M. Crenshaw, S. Croft, K. V. Croxall, E. Dalla Bontà,A. J. Deason, K. D. Denney, A. De Lorenzo-Cáceres, G. De Rosa, M. Dietrich, R. Edelson, J. Ely, M. Eracleous, P. A. Evans, M. M. Fausnaugh, G. J. Ferland, A. V. Filippenko, K. Flatland, O. D. Fox, E. L. Gates, N. Gehrels, S. Geier, J. M. Gelbord, V. Gorjian, J. E. Greene, C. J. Grier, D. Grupe, P. B. Hall, C. B. Henderson, S. Hicks, E. Holmbeck, T. W.-S. Holoien, D. Horenstein, K. Horne, T. Hutchison, M. Im, J. J. Jensen, C. A. Johnson, M. D. Joner, J. Jones, J. Kaastra, S. Kaspi, B. C. Kelly, P. L. Kelly, J. A. Kennea, M. Kim, S. Kim, S. C. Kim, A. King, S. A. Klimanov, C. S. Kochanek, K. T. Korista, G. A. Kriss, M. W. Lau, J. C. Lee, D. C. Leonard, M. Li, P. Lira, Z. Ma, F. MacInnis, E. R. Manne-Nicholas, M. A. Malkan, J. C. Mauerhan, R. McGurk, I. M. McHardy, C. Montouri, L. Morelli, A. Mosquera, D. Mudd, F. Muller-Sanchez, R. Musso, S. V. Nazarov, H. Netzer, M. L. Nguyen, R. P. Norris, J. A. Nousek, P. Ochner, D. N. Okhmat, B. Ou-Yang, A. Pancoast, I. Papadakis, J. R. Parks, L. Pei, B. M. Peterson, A. Pizzella, R. Poleski, J.-U. Pott, S. E. Rafter, H.-W. Rix, J. Runnoe, D. A. Saylor, J. S. Schimoia, K. Schnülle,S. G. Sergeev, B. J. Shappee, I. Shivvers, M. Siegel, G. V. Simonian, A. Siviero, A. Skielboe, G. Somers, M. Spencer, D. Starkey, D. J. Stevens, H.-I. Sung, J. Tayar, N. Tejos, C. S. Turner, P. Uttley, J. Van Saders, M. Vestergaard, L. Vican, S. Villanueva Jr., C. Villforth, Y. Weiss, J.-H. Woo, H. Yan, S. Young, H. Yuk, W. Zheng, W. Zhu, and Y. Zu, ”Space Telescope and Optical Reverberation Mapping Project. VII. Understanding the Ultraviolet Anomaly in NGC 5548 with X-Ray Spectroscopy”, The Astrophysical Journal, 846, 55, (2017)

14. M. M. Fausnaugh, C. J. Grier, M. C. Bentz, K. D. Denney, G. De Rosa, B. M. Peterson, C. S. Kochanek, R. W. Pogge, S. M. Adams, A. J. Barth, T. G. Beatty, A. Bhattacharjee, G. A. Borman, T. A. Boroson, M. C. Bottorﬀ, J. E. Brown, J. S. Brown, M. S. Brotherton, C. T. Coker, S. M. Crawford, K. V. Croxall, S. Eftekharzadeh, M. Eracleous, M. D. Joner, C. B. Henderson, T. W.-S. Holoien, K. Horne, T. Hutchison, S. Kaspi, S. Kim, A. L. King, M. Li, C. Lochhaas, Z. Ma, F. MacInnis, E. R. Manne-Nicholas, M. Mason, C. Montuori, A. Mosquera, D. Mudd, R. Musso, S. V. Nazarov, M. L. Nguyen, D. N. Okhmat, C. A. Onken, B. Ou-Yang, A. Pancoast, L. Pei, M. T. Penny, R. Poleski, S. Rafter, E. Romero-Colmenero, J. Runnoe, D. J. Sand, J. S. Schimoia, S. G. Sergeev, B. J. Shappee, G. V. Simonian, G. Somers, M. Spencer, D. A. Starkey, D. J. Stevens, J. Tayar, T. Treu, S. Valenti, J. Van Saders, S. Villanueva Jr., C. Villforth, Y. Weiss, H. Winkler, and W. Zhu, ”Reverberation Mapping of Optical Emission Lines in Five Active Galaxies”, The Astrophysical Journal, 840, 97, (2017)

15. L. Pei, M. M. Fausnaugh, A. J. Barth, B. M. Peterson, M. C. Bentz, G. De Rosa, K. D. Denney, M. R. Goad, C. S. Kochanek, K. T. Korista, G. A. Kriss, R. W. Pogge, V. N. Bennert, M. Brotherton, K. I. Clubb, E. Dalla Bont`a,A. V.

xv Filippenko, J. E. Greene, C. J. Grier, M. Vestergaard, W. Zheng, S. M. Adams, T. G. Beatty, A. Bigley, J. E. Brown, J. S. Brown, G. Canalizo, J. M. Comerford, C. T. Coker, E. M. Corsini, S. Croft, K. V. Croxall, A. J. Deason, M. Eracleous, O. D. Fox, E. L. Gates, C. B. Henderson, E. Holmbeck, T. W.-S. Holoien, J. J. Jensen, C. A. Johnson, P. L. Kelly, S. Kim, A. King, M. W. Lau, M. Li, C. Lochhaas, Z. Ma, E. R. Manne-Nicholas, J. C. Mauerhan, M. A. Malkan, R. McGurk, L. Morelli, A. Mosquera, D. Mudd, F. Muller Sanchez, M. L. Nguyen, P. Ochner, B. Ou-Yang, A. Pancoast, M. T. Penny, A. Pizzella, R. Poleski, J. Runnoe, B. Scott, J. S. Schimoia, B. J. Shappee, I. Shivvers, G. V. Simonian, A. Siviero, G. Somers, D. J. Stevens, M. A. Strauss, J. Tayar, N. Tejos, T. Treu, J. Van Saders, L. Vican, S. Villanueva Jr., H. Yuk, N. L. Zakamska, W. Zhu, M. D. Anderson, P. Arévalo, C. Bazhaw, S. Bisogni, G. A. Borman, M. C. Bottorff, W. N. Brandt, A. A. Breeveld, E. M. Cackett, M. T. Carini, D. M. Crenshaw, A. De Lorenzo-Cáceres,M. Dietrich, R. Edelson, N. V. Efimova, J. Ely, P. A. Evans, G. J. Ferland, K. Flatland, N. Gehrels, S. Geier, J. M. Gelbord, D. Grupe, A. Gupta, P. B. Hall, S. Hicks, D. Horenstein, K. Horne, T. Hutchison, M. Im, M. D. Joner, J. Jones, J. Kaastra, S. Kaspi, B. C. Kelly, J. A. Kennea, M. Kim, S. C. Kim, S. A. Klimanov, J. C. Lee, D. C. Leonard, P. Lira, F. MacInnis, S. Mathur, I. M. McHardy, C. Montouri, R. Musso, S. V. Nazarov, H. Netzer, R. P. Norris, J. A. Nousek, D. N. Okhmat, I. Papadakis, J. R. Parks, J.-U. Pott, S. E. Rafter, H.-W. Rix, D. A. Saylor, K. Schnülle,S. G. Sergeev, M. Siegel, A. Skielboe, M. Spencer, D. Starkey, H.-I. Sung, K. G. Teems, C. S. Turner, P. Uttley, C. Villforth, Y. Weiss, J.-H. Woo, H. Yan, S. Young, and Y. Zu, ”Space Telescope and Optical Reverberation Mapping Project. V. Optical Spectroscopic Campaign and Emission-line Analysis for NGC 5548”, The Astrophysical Journal, 837, 131, (2017)

16. S. Villanueva Jr., J. D. Eastman, B. S. Gaudi, R. W. Pogge, K. G. Stassun, M. Trueblood, and P. Trueblood, ”DEdicated MONitor of EXotransits and Transients (DEMONEXT): a low-cost robotic and automated telescope for followup of exoplanetary transits and other transient events”, Ground-based and Airborne Telescopes VI, 9906, 99062L, (2016)

17. S. Villanueva Jr., J. D. Eastman, and B. S. Gaudi, ”The Dedicated Monitor of Exotransits (DEMONEX): Seven Transits of XO-4b”, The Astrophysical Journal, 820, 87, (2016)

18. J. L. Marshall, D. L. DePoy, T. Prochaska, R. D. Allen, P. Williams, J.- P. Rheault, T. Li, D. Q. Nagasawa, C. Akers, D. Baker, E. Boster, C. Campbell, E. Cook, A. Elder, A. Gary, J. Glover, M. James, E. Martin, W. Meador, N. Mondrik, M. Rodriguez-Patino, S. Villanueva, G. J. Hill, S. Tuttle, B. Vattiat, H. Lee, T. S. Chonis, G. B. Dalton, and M. Tacon, ”VIRUS instrument collimator assembly”, Ground-based and Airborne Instrumentation for Astronomy V, 9147, 91473S, (2014)

xvi 19. S. Villanueva Jr., D. L. DePoy, and J. L. Marshall, ”Optimal resolutions for optical and NIR spectroscopy”, Ground-based and Airborne Instrumentation for Astronomy IV, 8446, 84462V, (2012)

20. D. L. DePoy, R. Allen, R. Barkhouser, E. Boster, D. Carona, A. Hard- ing, R. Hammond, J. L. Marshall, J. Orndorﬀ, C. Papovich, K. Prochaska, T. Prochaska, J. P. Rheault, S. Smee, S. Shectman, and S. Villanueva, ”GMACS: a wide ﬁeld, multi-object, moderate-resolution, optical spectrograph for the Giant Magellan Telescope”, Ground-based and Airborne Instrumentation for Astronomy IV, 8446, 84461N, (2012)

21. J. E. Thomas-Osip, G. Prieto, A. Berdja, K. W. Cook, S. Villanueva Jr., D. L. DePoy, J. L. Marshall, J. P. Rheault, R. D. Allen, and D. W. Carona, ”Characterizing Optical Turbulence at the GMT Site with MooSci and MASS- DIMM”, Publications of the Astronomical Society of the Paciﬁc, 124, 84, (2012)

22. M. D. Stritzinger, M. M. Phillips, L. N. Boldt, C. Burns, A. Campillay, C. Contreras, S. Gonzalez, G. Folatelli, N. Morrell, W. Krzeminski, M. Roth, F. Salgado, D. L. DePoy, M. Hamuy, W. L. Freedman, B. F. Madore, J. L. Marshall, S. E. Persson, J.-P. Rheault, N. B. Suntzeﬀ, S. Villanueva, W. Li, and A. V. Filippenko, ”The Carnegie Supernova Project: Second Photometry Data Release of Low-redshift Type Ia Supernovae”, The Astronomical Journal, 142, 156, (2011)

23. A. D. Collins, B. Vattiat, J. L. Marshall, G. J. Hill, D. L. DePoy, H. Lee, R. D. Allen, T. Prochaska, and S. Villanueva Jr., ”Development of VIRUS alignment and assembly ﬁxtures”, Ground-based and Airborne Instrumentation for Astronomy III, 7735, 773574, (2010)

24. S. Villanueva Jr., D. L. Depoy, J. Marshall, A. Berdja, J. P. Rheault, G. Prieto, R. Allen, and D. Carona, ”MooSci: a lunar scintillometer”, Ground-based and Airborne Instrumentation for Astronomy III, 7735, 773547, (2010)

Fields of Study

Major Field: Astronomy

xvii Table of Contents

Abstract ...... ii

Dedication ...... v

Acknowledgments ...... vi

Vita ...... xi

List of Tables ...... xxii

List of Figures ...... xxiii

Chapter 1: Introduction ...... 1

1.1 A Brief History of Exoplanet Discoveries ...... 1

1.2 The State of Exoplanet Discovery ...... 3

1.3 Understanding Hot Jupiters ...... 4

1.4 Scope of the Dissertation ...... 5

Chapter 2: Analysis of the Known Hot-Jupiter XO-4b ...... 7

2.1 Introduction ...... 7

2.2 Data ...... 8

2.2.1 DEMONEX Observations of XO-4 ...... 9

2.2.2 McCullough 2008 ...... 10

xviii 2.2.3 Narita 2010 ...... 11

2.2.4 Todorov 2012 ...... 11

2.3 Models ...... 12

2.3.1 Transits and Radial Velocity ...... 12

2.3.2 Rossiter-McLaughlin Eﬀect ...... 12

2.4 Data Analysis ...... 17

2.4.1 Narita Data vs Narita 2010 ...... 17

2.4.2 DEMONEX Data ...... 19

2.4.3 Combined Data Set ...... 22

2.5 Results ...... 28

2.6 Chapter Summary ...... 29

Chapter 3: DEMONEXT ...... 51

3.1 Introduction ...... 51

3.2 Science Drivers ...... 52

3.2.1 KELT Exoplanet Follow-Up ...... 52

3.2.2 ASAS-SN Transient Follow-Up ...... 54

3.2.3 Ancillary Science ...... 55

3.2.4 Summary of Science Requirements ...... 55

3.3 Optical and Mechanical System ...... 56

3.3.1 Optical Telescope Assembly and Mount ...... 56

3.3.2 Camera, Filter, and Focuser ...... 57

3.3.3 Guide Scope ...... 57

3.3.4 Remote Power and Connectivity ...... 58

3.4 Nightly Operations ...... 59

xix 3.4.1 Software ...... 59

3.4.2 System Initialization ...... 60

3.4.3 Nightly Calibrations ...... 60

3.4.4 Observations ...... 61

3.4.5 Continuous Time-Series Observations ...... 61

3.4.6 Long-Term Monitoring ...... 63

3.5 Data Processing ...... 64

3.5.1 Bias, Dark, and Flat Corrections ...... 64

3.5.2 World Coordinate Solution ...... 64

3.5.3 Reductions Left to Users ...... 65

3.6 System Performance ...... 65

3.6.1 Noise ...... 65

3.6.2 Guiding ...... 67

3.6.3 Automation Yields ...... 70

3.7 DEMONEXT Science Results ...... 71

3.7.1 KELT Exoplanet Follow-Up ...... 71

3.7.2 ASAS-SN Monitoring and Conﬁrmation ...... 73

3.7.3 Microlensing Surveys and Follow-Up ...... 74

3.7.4 Open Clusters ...... 75

3.8 Chapter Summary ...... 76

Chapter 4: Single Transits in TESS ...... 91

4.1 Introduction ...... 91

4.2 Expected Number of Single-Transit Planets ...... 93

4.2.1 Demographics of Detected Single-Transits ...... 98

xx 4.3 Estimating the Period ...... 100

4.3.1 Uncertainty on the Period Due to the Stellar Density . . . . . 101

4.3.2 Uncertainty on the Period Due to the Photometry ...... 103

4.3.3 Uncertainty on the Period Due to Eccentricity ...... 104

4.4 Prospects for Follow-up ...... 105

4.4.1 Recovery with Additional Photometry or Precovery in Archival Data ...... 105

4.4.2 Expected Radial Velocity Signal ...... 105

4.5 Comparison to Other Simulations ...... 106

4.6 Recommendations for Observations ...... 107

4.7 Chapter Summary ...... 108

Chapter 5: Conclusions and Future Work ...... 125

5.1 Summary ...... 125

5.2 Future Work ...... 125

References ...... 126

xxi List of Tables

Table 2.1 XO-4b values compared to the literature...... 44

Table 2.2 DEMONEX sample data including trend parameters for XO-4b.b 45

Table 2.3 Transit Parameters of XO-4b with DEMONEX...... 46

Table 2.4 Transit Times for XO-4b ...... 47

Table 2.5 Median values and 68% conﬁdence interval for XO-4b...... 48

Table 2.5 Median values and 68% conﬁdence interval for XO-4b...... 49

Table 2.5 Median values and 68% conﬁdence interval for XO-4b...... 50

Table 3.1 Summary of DEMONEXT parts and components...... 89

Table 3.2 Summary of DEMONEXT software...... 90

Table 4.1 Fraction of sky covered by various observing baselines...... 124

xxii List of Figures

Figure 2.1 Seven DEMONEX light curves of XO-4b...... 31

Figure 2.2 Binned DEMONEX data of XO-4b...... 32

Figure 2.3 DEMONEX detrend parameters from XO-4b...... 33

Figure 2.4 DEMONEXT TTVs from detrended parameters of XO-4b. . . 34

Figure 2.5 Archival light curves of XO-4b...... 35

Figure 2.6 Global binned light curves of XO-4b...... 36

Figure 2.7 Allan Variance of XO-4b...... 37

Figure 2.8 Residual Uncertainties of XO-4b...... 38

Figure 2.9 Radial Velocity Curve of XO-4b...... 39

Figure 2.10 Rossiter-McLauglin of XO-4b...... 40

Figure 2.11 Spin-Orbit (Miss)Alignment of XO-4b...... 41

Figure 2.12 20171103 Light Curve of XO-4b...... 42

Figure 2.13 Transit Timing Variations of XO-4b...... 43

Figure 3.1 DEMONEXT telescope ...... 77

Figure 3.2 DEMONEXT Precision ...... 78

Figure 3.3 Allen Variance ...... 79

Figure 3.4 DEMONEXT Uncertainties ...... 80

Figure 3.5 DEMONEXT Science Guiding ...... 81

Figure 3.6 DEMONEXT Yields ...... 82

xxiii Figure 3.7 KELT-20b ...... 83

Figure 3.8 Supernova 2016hli ...... 84

Figure 3.9 ASASSN-16fm ...... 85

Figure 3.10 DEMONEXT Mirco-Lensing Campaign ...... 86

Figure 3.11 DEMONEXT Mirco-Lensing Events ...... 87

Figure 3.12 DEMONEXT Stellar Rotation ...... 88

Figure 4.1 Probability of a Single Transit ...... 110

Figure 4.2 Mission Weighted Probability of a Single Transit ...... 111

Figure 4.3 Planet Occurance Rates ...... 112

Figure 4.4 TESS Expected Yield ...... 113

Figure 4.5 TESS Demographics: Magnitude ...... 114

Figure 4.6 TESS Demographics: Eﬀective Temperature ...... 115

Figure 4.7 TESS Demographics: Planet Radius ...... 116

Figure 4.8 TESS Demographics: Stellar Insulation ...... 117

Figure 4.9 TESS Demographics: Signal-to-Noise ...... 118

Figure 4.10 Photometric Uncertainty from Photometry ...... 119

Figure 4.11 Eccentricity Eﬀects ...... 120

Figure 4.12 TESS Demographics: Transit Depths ...... 121

Figure 4.13 TESS Demographics: Radial Velocity Signal ...... 122

Figure 4.14 TESS Yield for Multi-transit Events ...... 123

xxiv Chapter 1: Introduction

1.1. A Brief History of Exoplanet Discoveries

The field of exoplanets has a long history checkered with many claimed exoplanets later proven to be instrumental effects or false positives. Using the radial velocity (RV) technique to measure the mass and period of a planet, the first claimed RV planets came in the late 1980’s (Campbell et al. 1988; Latham et al. 1989), but were ambiguous at the time. These were followed by the radial-timing detection of two planets orbiting the pulsar PSR B1257+12 (Wolszczan & Frail 1992). The field changed dramatically in 1995 following the first detection of an unambiguously planetary mass object (0.47MJ) orbiting the main sequence star 51 Pegasi on a 4.2 day orbit (Mayor & Queloz 1995). The discovery of the first Hot Jupiter opposed the notion that other solar systems would follow the same architecture as our own. The assumption was that giant planets are expected to reside at large separations beyond a few astronomical units (AU), while terrestrial planets occupy the interior at less than a few AU. The planet 51 Peg b, and the other Hot Jupiters that would follow, challenged the known models and theories of planet formation, disk formation, and planet migration (Marcy et al. 2005; Fabrycky & Tremaine 2007; Nagasawa et al. 2008). Following 51 Peg b, the number of radial velocity detected planets steadily increased towards the turn of the century as radial velocity surveys established longer baselines and became sensitive to planets on larger orbits. Radial velocity studies

1 provided the ﬁrst insights into the nature of the distribution of planets, including estimates of the mass distributions of exoplanets and their orbital properties.

The early years of exoplanets discoveries were not limited to radial velocity detections. In 1999 the first detection of a transit from a radial velocity discovered planet opened the doors to the transit method as a viable tool to probe the geometric nature of exoplanets (Henry et al. 1999; Charbonneau et al. 2000). Transits allowed for the first measurements of the radii of exoplanets, in particular those at small orbits where the probability to observe a transit increases. The transit method is most sensitive to the largest planets around the smallest stars on the shortest orbits, and is thus a viable method for understandin Hot Jupiters. These discoveries prompted a number of photometric surveys and the first planet discovered from a transit was found in the OGLE survey in 2002 (Konacki et al. 2003).

The first decade of exoplanet discovery provided the first results in a number of key detection and characterization methods: radial velocity, timing, transit, microlensing (Bond et al. 2004), and astrometry(Benedict et al. 2006). Transits aside, these methods are particularly useful at provided constraints on the masses of planets and their orbits, while being sensitive to the largest planets on tight (RV) and wide (mircolensing, direct imaging) orbits, around the smallest stars (microlensing and timing are the exceptions with sensitivities to low-mass planets). The second decade saw an exponential growth in the development of the field. A greater number of resources have been devoted, both in funding and telescope time, towards the pursuit of a true Earth analog. The “race to the bottom” is pushing the sensitivities for all available methods to the lowest masses possible. This has driven a number of technological and instrumental developments in the field. Many regard the detection of an Earth-mass planet with an Earth-like radius on an Earth-like orbit around a Sun-like star as the “Holy Grail” of exoplanet discoveries, and the field has benefited

2 from the instrumental developments that have enabled an exponential growth in discoveries.

1.2. The State of Exoplanet Discovery

With an increase in technological performance, radial velocity discoveries have increased with the advent of 1 m s−1 precision spectrographs. The orders of magnitude improvement in contrast from adaptive optics systems and coronagraphs are providing the ability to directly image planets on wide orbits, including multiple planet systems (Lagrange et al. 2009). Combine these with the 40 microlensing detections to date1 and you have an ever growing sample of long-period giant planets. The development of ground-based, large area, bright star photometric surveys such as HAT, Tres, XO, WASP, and KELT (Bakos et al. 2004; Alonso et al. 2004; McCullough et al. 2005; Pollacco et al. 2006; Pepper et al. 2007) have yielded hundreds of planet detections in the past decade, and provide key insights into understanding Hot Jupiters. The various approaches of these surveys have yielded a variety of planets, but all are limited to a 1–0.1% precision due to the limitations of observing from the ground. The Tres, XO, HAT, and WASP surveys have monitored stars of 12-16th magnitude, resulting in an abundance of short-period planet detections. Other surveys have been more targeted. The KELT survey is targeting the brightest stars in they sky and is ﬁnding some of the hottest planet hosting stars known. Surveys like MEarth are performing targeted studies of the smaller M dwarfs, where the total number of targets are decreased, but the chances of ﬁnding the smallest planets from the ground are increased (Irwin et al. 2009). In total the hundreds of ground-based transits have provide insight into the demographics of the Hot Jupiters and other short period planets. The launch of

1exoplanets.eu

3 the CoRoT and speciﬁcally Kepler space-based telescopes have given the transit community the ability to detect the smallest transiting objects to date (Auvergneet et al. 2009; Borucki et al. 2010). The increased performance from space and better understanding of systematic errors, or lack thereof, have enabled the 10−4 precision in transit light curves required to detect Earth analogs. Kepler is one of the few resources sensitive to terrestrial planets and Kepler sets the occurrence rate for terrestrial planets to date (Fressin et al. 2013).

1.3. Understanding Hot Jupiters

To understand how Hot Jupiters form and evolve, we must understand the planets’ properties, especially masses and radii, distributed over a range of orbits, with a statistically significant number of planets. From this sample, a sub-sample of optimal targets are then selected for additional characterization to understand atmospheric composition and planetary structure. We currently have a growing number of Hot Jupiters, but we do not have a large or diverse sample of planets with both accurate masses and radii, distributed over a broad range of orbits that probe a broad range of stellar irradiation. As a result, we are unable to disentangle the effects of stellar irradiation on Hot Jupiter properties from the effects of planet formation and migration.

What we can do, is measure the known Hot Jupiters in situ. A planet’s mass and radius is required to measure its bulk density. This is the first step in understanding the nature of a planet, whether it be a dense terrestrial (i.e., rocky) planet like Earth, or a gaseous giant like Jupiter or Saturn. Although a planet’s bulk density is a blunt tool for classifying a planet, it places many constraints on planet formation models. However, it is known that planets that receive high incident fluxes from their host stars, have radii that are substantially inflated above theoretical

4 models (Demory & Seager 2011; Thorngren et al. 2016). To disentangle the effects of formation and migration from evaporation and inflation requires the detection and confirmation of planets over a wide range of orbits, including orbits that are large enough that they are unlikely to affected by stellar irradiation. As such, we will eventually want to study long-period Hot Jupiters, or Warm Jupiters.

1.4. Scope of the Dissertation

The outline of this dissertation is as follows: In Chapter 2, I discuss the analysis of a known transiting Hot Jupiter, XO-4b, as featured in Villanueva et al. (2016a). This data set includes new data from the DEdicated MONitor of EXotransits (DEMONEX) telescope. We combine these data with archival light curves and archival radial velocity measurements to derive the properties of the host star and plane, including archival Rossiter-McLaughlin measurements. This chapter also tests the eﬀects of including various detrend parameters, theoretical and empirical mass-radius relations, and Rossiter-McLaughlin models on the inferred parameters. We comment on the eﬀects of detrending the time-series photmetry can have data quality and the inferred values of many parameters.

In Chapter 3 I report on the design and first years of operations of the DEdicated MONitor of EXotransits and Transients (DEMONEXT) as featured in Villanueva et al. (2018a). DEMONEXT is the successor to DEMONEX, and is a 20 inch (0.5-m) robotic telescope using a PlaneWave telescope on a Mathis instruments fork mount. DEMONEXT is equipped with a CCD detector, filter wheel, electronic focuser, and nine standard astronomy filters. We discuss the observing modes created for the telescope, the performance of the time-series photometry, and the system stability. We highlight the efficiency and data products through the automation

5 routines, and discuss a number of scientiﬁc highlights from the ﬁrst few years of DEMONEXT operations.

In Chapter 4 I present a semi-analytic estimate of the expected yield of single-transit planets from the Transiting Exoplanet Survey Satellite (TESS), a submitted manuscript (Villanueva et al. 2018b). I use the TESS Candidate Target List 6 (CTL-6) as an input catalog of over 4 million sources to predict the number of single transits expected to be found from 200,000 target stars and in the full-frame images (FFIs). This includes demographics of the expected planets and the ability to observe these planets after TESS with observatories such as DEMONEXT.

In Chapter 5 I provide conclusions and potential avenues for future work.

6 Chapter 2: Analysis of the Known Hot-Jupiter XO-4b

This chapter is based on work published in Villanueva et al. (2016a).

2.1. Introduction

Hot Jupiters are an unique class of extrasolar planets (exoplanets) known for their proximity to their host stars and their large masses. With the discovery of the ﬁrst Hot Jupiter (Mayor & Queloz 1995), we now know that understanding the physical nature of Hot Jupiters can play a large role in constraining our understanding of a variety of theories regarding planetary formation, disk formation, and planet migration (Marcy et al. 2005; Fabrycky & Tremaine 2007; Nagasawa et al. 2008).

Hot Jupiters provide unique observational opportunities relative to exoplanets of other populations. The small semi-major axis of the orbits of Hot Jupiters increase the a priori transit probability, this coupled with a large observational transit signal due to Hot Jupiters’ large relative radii, and their short periods provide many observational opportunities upon which to identify Hot Jupiters and perform follow-up observations. These observational advantages are not unique to photometric observations, but also to radial velocity (RV) measurements where the large masses of Hot Jupiters increase the semi-amplitude of the radial velocity signal as well as to Rossiter-McLaughlin (RM) measurements where the RM signal is proportional to the transit depth.

7 In general, the transits provide a number of geometric ratios to relate the star and planet (i.e., the depth of the transit is related to the planet/star radius ratio), as well as the density of the star. When combined with another constraint on the mass and radius of the primary (e.g., derived from isochrones using the spectroscopic stellar eﬀective temperature and metallicity), and a measurement of the Doppler amplitude from radial velocity observations, it becomes possible to estimate the masses and radii of both the planet and star. Additional measurements allow for the derivation of a large number of physical parameters, not just the planet and stellar masses and radii (see Winn 2010).

Following the discovery of systems containing Hot Jupiters, one generally would like to follow-up with observations to better understand the properties of the host star and planet. In general this requires additional photometric and radial velocity observations. Providing stricter constraints on known Hot Jupiters provides predictions for future observations, increasing the eﬃciency of those observations.

The purpose of this paper is three-fold in nature. First, to present new data on XO-4b from the DEdicated MONitor of EXotransits (DEMONEX) telescope. Second, we investigate the effects of choices of detrending variables, mass-radius relations, Rossiter-McLaughlin models, and other models or user defined variables on the derived parameters of the system. Third, we improve the parameters of XO-4 and XO-4b by globally fitting all available data in a homogeneous manner.

2.2. Data

The following sections contain summaries of all available data used in our global analysis as well as information regarding the data reduction process used for the new DEMONEX data.

8 2.2.1. DEMONEX Observations of XO-4

New observations of XO-4b were made using the DEdicated MONitor of EXotransits (DEMONEX) (Eastman et al. 2010a). DEMONEX was a low-cost, 0.5 meter, robotic telescope constructed from commercially-available parts operated remotely out of Winer Observatory in Sonoita, Arizona. DEMONEX monitored bright stars hosting known transiting planets over a three year period from 2008-2011 in order to provide a homogeneous data set of precise relative photometry for over 40 transiting systems. The DEMONEX mount became increasingly unreliable and failed on multiple occasions requiring both major and minor repairs. After three years of operations DEMONEX was decommissioned due to the the maintenance costs and resources required to keep the mount operational from its remote location.

There are 20 nights of data from November 2008 to May 2010 taken during primary transits of XO-4b. All observations were made in the Sloan z band. Due to issues with the mount, 5 nights were lost due to the mount pointing to the incorrect field. In addition, the DEMONEX observing strategy prioritized full transits of other targets over partial transits of XO-4b. Often, that resulted in small observing windows, which we opted to fill with observations near transit in case they happened to be useful rather than sitting idle. Unfortunately, 4 nights only got data out-of-transit and 4 nights only captured the flat bottom of the transit, neither of which ended up providing useful constraints. Thus we were left with an unfortunately low yield of 7/20 (35%) usable nights, but we obtain 3 full transits, 2 ingresses, 2 egresses.

9 2.2.2. McCullough 2008

McCullough et al. (2008) (MC08) report the original discovery of the planet XO-4b detected using the XO telescope (McCullough et al. 2006). Follow-up BVR observations were made by the XO Extended Team (XOET) as well as follow-up R band photometric observations using the Perkins 1.8-m telescope. Follow-up spectroscopic measurements were made using the Harlan J. Smith 2.7-m telescope and the 11-m Hobby-Eberly Telescope. MC08 perform an analysis of the spectra to report the stellar properties of the host star including stellar eﬀective temperature Teﬀ , surface gravity log g∗, metallicity [Fe/H], projected rotational velocity v sin I∗, and the RV semi-amplitude K. Combining the stellar parameters with the light curves MC08 report a planet mass of MP = 1.72 ± 0.20 MJ , radius of

RP = 1.34 ± 0.048 RJ , orbital period of P = 4.12502 ± 0.00002 days, and heliocentric Julian date at mid-transit of 2454485.9322 ± 0.0004 for XO-4b.

We adopt the stellar properties, Teﬀ , log g∗, [Fe/H], and v sin I∗, from MC08 as Gaussian priors for our global analysis as described in §2.4. Additionally, the authors of MC08 kindly provided the original XO light curves, the BVR band XOET follow-up light curves, and the Perkins follow-up light curves, which we use, along with the radial velocity data listed in the MC08 paper in our global analysis.

We convert the times from the MC08 XOET light curves from HJDUT to

1 BJDTDB using the IDL code HJD2BJD to maintain uniform time stamps across all available light curves. The average correction is ∼ 94 seconds, which is less than the typical uncertainty in the central transit times.

1http://astroutils.astronomy.ohio-state.edu/time/hjd2bjd.html

10 2.2.3. Narita 2010

Narita et al. (2010) (N10) report new Sloan z band photometric and radial velocity observations of XO-4 conducted with the FLWO 1.2 m telescope (photometric) and the 8.2 m Subaru Telescope (RV). Based on these new light curves, N10 report a reﬁned transit ephemeris for XO-4b of P = 4.1250828 ± 0.0000040 days and

Tc = 2454485.93323 ± 0.00039 BJDTDB.

N10 also report the ﬁrst measurements of the Rossiter-McLaughlin eﬀect of XO-4b. N10 estimate the sky-projected angle between the stellar spin axis an the

+8.1 planetary orbital axis to be λ = −46.7−6.1 degrees. We compare the N10 results from the publicly-available N10 light curves and radial velocity data against the results we obtain using the same data set outlined in §2.4.1 to validate our light curve analysis methods and procedures. We also include the publicly-available N10 light curves and radial velocity data in our global analysis.

2.2.4. Todorov 2012

Todorov et al. (2012) use the Spitzer Space Telescope (Werner et al. 2004) and Infrared Array Camera (Fazio et al. 2004) to obtain 3.6 µm and 4.5 µm observations of the secondary eclipses of three planets including XO-4b. To better constrain the ephemerides of the planetary transits Todorov et al. (2012) make additional Ic band ground based primary transit observations using the Universidad de Monterrey Observatory (UDEM) 0.36-m telescope. We do not include secondary eclipse data in our data analysis, but we do include the UDEM Ic band ground based primary transit observations in our global analysis, including the previously unpublished UT 2012-01-07 data, kindly provided by the authors.

11 We convert the times from the Todorov et al. (2012) UDEM observations from

BJDUT to BJDTDB using the IDL code JDUTC2JDTDB to maintain uniform time stamps across all available light curves. While this routine is intended to convert Geocentric Julian Date from UTC to TDB, it is accurate to 30 ms when using it to convert Barycentric Julian Dates instead (Eastman et al. 2010b), which is more than suﬃcient for our purposes. The average correction is ∼ 82 seconds, which is less than the typical uncertainty in the central transit times.

2.3. Models

2.3.1. Transits and Radial Velocity

The models used in this paper to ﬁt the transit and orbital radial velocity data are unchanged from the original Eastman et al. (2013) EXOFAST paper. The original EXOFAST paper ignores a number of eﬀects, including RM and transit timing variations (TTVs), that we now include to maintain consistency with work done by the other groups.

2.3.2. Rossiter-McLaughlin Eﬀect

We also consider the radial velocity data taken during transit, and as such must consider models for the RM effect (Rossiter 1924; McLaughlin 1924). The precise model used to model the RM effect can have a significant effect on the inferred parameters, and yet this can vary from system to system (Johnson et al. 2008; Hirano et al. 2010). A number of RM models exist, and we investigate two separate models here.

The ambiguity of the proper model of the RM eﬀect that should be used comes from the fact that the RM eﬀect is not, in fact, due to a change in the radial velocity

12 of the star. The radial velocity measurement is made by estimating some measure of the “centers” of known absorption lines and their change in central wavelength relative to laboratory measurements. If the shape of the absorption features changes in such a way as to change the odd moments of the lines, then this can result in a change in the measured line centers, and as a result be attributed to a change in the radial velocity of the star. The precise relation between the change in the line shape and the inferred change in the line center will depend on many intrinsic properties of the star, such as its eﬀective temperature, surface gravity, and rotation rate, as well as the precise algorithm used to estimate the line centers.

Thus the RM eﬀect manifests as an anomalous radial velocity measurement

∆vRM made during the primary transit, where the anomalous signal is due to a change in the shape of the absorption feature rather than to the motion of the host star. The change in shape is due to the transiting planet preferentially blocking light emitted from the rotating host star. The blocked light, depending on the position of the planet as viewed from the observer, may be red or blue shifted relative to the center of the star due to the rotation of the host star. As this light is blocked it can introduce an asymmetry to the absorption feature (Gaudi & Winn 2007). Because the anomalous signal is dependent on a change in the shape of the absorption feature, the measured ∆vRM will depend heavily on the method used to measure the radial velocity signal, and whether that method is sensitive to changes in the shape of the absorption features.

The RM anomalous radial velocity shift has a strong dependence on the host star’s rotation v sin I∗ and the axis of the orbit of the planet relative to the sky-projected spin axis of the star λ. The RM signal is also dependent on the radius of the planet in stellar radii Rp/R∗ = p, the limb-darkening parameters u1 and u2, and the path the planet takes over star which can be described by the

13 inclination i and orbital distance a or by the impact parameter b. Winn (2010) gives an approximation for the maximum amplitude of the RM eﬀect as:

√ 2 2 ∆vRM ≈ p 1 − b (v sin I∗) (2.1)

When the eﬀect is observed it is possible to place constraints on the sky-projected spin axis and planetary orbit axis alignment λ, sky-projected rotational velocity v sin I∗, and the impact parameter b. The transit light curves and previous RV studies provide us with independent constraints on v sin I∗ and b, and these improve the constraint on λ.

There are two different RM models we investigate to place the constraint on λ, those based on the moment method (Ohta et al. 2005, 2009) (OTS) and those based on the cross-correlation method (Queloz et al. 2000; Winn et al. 2005; Narita et al. 2009; Hirano et al. 2010). There are other models available (Giménez2006; Bouéet al. 2013), but we focus on the OTS and cross-correlation methods as the default model in EXOFAST is that based on Ohta et al. (2005) and previous work on XO-4b follows the cross-correlation method. Additional constraints from Doppler tomography (Collier Cameron et al. 2010; Bourrier et al. 2015) could be used to provide an independent check on the accuracy of the two methods.

Moment Method

Ohta et al. (2005) describe their derivation of the RM effect as an approximate but accurate analytic formula for the anomaly in the radial velocity curves. Their method approximates the velocity anomaly as the change in the first moment of the absorption line profile and uses only linear stellar limb-darkening. This work is followed by Ohta et al. (2009) where the authors present theoretical predictions for the photometric and spectroscopic signatures of rings around transiting extrasolar

14 planets that now include quadratic limb-darkening and terms for ringed extrasolar planets. These new expressions supersede the work in Ohta et al. (2005).

OTS deﬁne the ﬂux of the star as a function of the position of the planet

RR I(x, z)dxdz F = RR , (2.2) I∗(x, z)dxdz where x and z are the position of the center of the planet perpendicular to and parallel to, respectively, the projected rotation axis of the star, and they integrate over the surface brightness of the unocculted stellar disk I∗, and I is the surface brightness of the occulted star, such that I = 0 in the region occulted by the planet. OTS assume rigid rotation in the star, i.e. no diﬀerential rotation such that the projected velocity is constant along lines of constant x.

When we use these expressions we ignore the eﬀects of planetary rings as they are not expected around Hot Jupiters (Gaudi et al. 2003). Under this assumption, the anomalous radial velocity can be reduced the following expression:

∆vRM = xpv sin I∗F (2.3)

where xp is the x component of the planet’s position in units of stellar radii and F is the relative ﬂux from Equation (2.2). The details of these expressions are included in Ohta et al. (2009), including detailed expressions for integrating the ﬂux in Equation (2.2).

Cross-Correlation Method

An alternative method to measure the RM eﬀect is to compare the anomalous radial velocity by cross-correlation with a stellar template spectrum. This method was used by Queloz et al. (2000), Winn et al. (2005), Narita et al. (2009), among others,

15 and is described in detail by Hirano et al. (2010). Hirano et al. (2010) cover in detail the choice of absorption line proﬁle (Gaussian, Voigt, etc.), and eventually settle on a form inspired by their Gaussian approximation:

2 ∆vRM = −F vp(p − qvp) (2.4)

Here F is the ﬂux ratio and uses the same expression as that of Equation (2.2). For rigidly rotating stars v(x, y) sin I∗ is a constant along lines of constant x and the expression reduces to

vp(x, y) = xv sin I∗, (2.5) and reproduces the OTS result as noted by Hirano et al. (2010).

The parameters p and q are each functions of thermal broadening, micro- turbulent broadening, and the stellar rotation width that describe the shape of the line. The parameters p and q are empirically ﬁt for each individual planetary system the method is applied to. As a result, to ﬁrst order, for which p = 1 and q = 0, Equation (2.4) reduces to Equation (2.3) and the two methods are identical. For reference, in the case of XO-4b, Narita et al. (2010) derive and report p = 1.6159 and q = 0.83778, but do not quote errors or uncertainties for these values.

Because p and q depend on precisely how the algorithm used to estimate the radial velocity measures the shape of the absorption profiles and the line spread function of the spectrograph, it is possible to confuse a change in line shape as instrumental and not due to the RM effect. If ignored, this can lead to a misinterpretation of the RM effect, and as such this method must be applied to each system as outlined above to distinguish between the two effects. There are cases when the moment method and cross-correlation methods produce the same results

16 (Winn et al. 2008; Johnson et al. 2008). However, discrepancies in the derived parameters can arise between the two methods, particularly for systems with small impact parameters (Winn et al. 2005; Hirano et al. 2011).

2.4. Data Analysis

To analyze the XO-4b data we use a custom version of the publicly available EXOFAST suite of IDL programs described in Eastman et al. (2013). This is a continuation of the same code used to fit the KELT discoveries (e.g., Siverd et al. 2012), and includes the ability to simultaneously fit multiple transit light curves, both on the same telescope from different nights and from different telescopes, as well as the ability to include multiple radial velocity data sets. Additional modifications were made by us to update the default RM models, include alternative RM models, and to change how the estimated errors are scaled on the RM data.

We adopt Gaussian priors from MC08 on the stellar parameters Teﬀ , log g∗,

[Fe/H], and v sin I∗, and use the reported values of p = 1.6159 and q = 0.83778 from N10 when using the Hirano et al. (2010) based RM models. We do ignore secondary eclipse data from Todorov et al. (2012) and assume e = 0 as this is consistent with all of the other groups (McCullough et al. 2008; Narita et al. 2010; Todorov et al. 2012).

2.4.1. Narita Data vs Narita 2010

As a conﬁrmation of our light curve analysis we make many of the same assumptions used in the N10 paper and compare the output from EXOFAST using the N10 data to the published results of the N10 paper. It should be noted that N10 include the same Gaussian priors we adopt, and additionally include a Gaussian prior on the period and T0 from MC08. N10 also ﬁx the limb-darkening parameter u1 and

17 treat the second limb-darkening parameter u2 as a free parameter. In EXOFAST both u1 and u2 are calculated from log g∗, Teff , [Fe/H], and the observed band-pass from Claret & Bloemen (2011). During this exercise we do not create a new free parameter for u2, nor do we fix u1, but we do include the Gaussian priors on period and T0. Thus our results are not precisely comparable, although we expect any differences to be relatively minor.

Table 2.1 contains the same parameters published in N10 (Naritaχ2 ) along with those produced by EXOFAST (EXOFASTMCMC). The values are consistent within the errors for all parameters. It should be noted that N10 use the minima in χ2 and ∆χ2 = 1 to quote their best value and errors, where we typically quote the median values of the parameters and the 68% conﬁdence intervals from the MCMC chains. In order to more precisely compare our results to those of N10, we also ﬁt a multi-dimensional hyperboloid to the output MCMC chains in order to infer the values of the parameters with the minimum χ2 and the values where ∆χ2 = 1

(EXOFASTχ2 ).

As we did not create a new free parameter for u2, the constraint on the second limb-darkening parameter is calculated from the parameters log g∗, Teff , [Fe/H], and the band. Thus the uncertainty on u2 is simply a reflection on the covariance of u2 with these parameters. We generally find values and uncertainties that are in good agreement with those of N10, verifying that the methods used in N10 are comparable to those used in EXOFAST. The final column of Table 2.1 shows the differences in the two minimum χ2 values divided by the uncertainties in quadrature for reference. The differences between methods are less than 1 σ and any differences may be attributable to our slightly different methods of estimating the values for the minimum χ2 and ∆χ2 = 1.

18 2.4.2. DEMONEX Data

We converted all times in the DEMONEX data to barycentric Julian date in the barycentric dynamical time (BJDTDB) as advocated in Eastman et al. (2010b). We performed standard data reduction procedures for the raw DEMONEX XO-4 data. These include bias correction, dark subtraction, flat fielding, and due to our Sloan z band observations we additionally perform fringe corrections. To perform the fringe corrections a master fringe image is created by taking the median of the nearest 100 images separated by a minimum time step, where the stars are masked out in each image. The background and amplitude of the master fringe image are fit and then subtracted off of each science image.

We use AstroImageJ2 (AIJ) to perform aperture photometry on the target star and a set of comparison stars to create normalized relative photometric light-curves for XO-4. We use an initial set of comparison stars based on their similar counts to XO-4. For each night the flux from the target star XO-4 was divided by the sum of the flux from the set of comparison stars in the image and normalized to unity. We then select the the comparison star that gives the lowest out-of-transit (pre-ingress and post-egress) root-mean-square (RMS) for XO-4, and we add additional comparison stars only if their inclusion reduces the out-of-transit RMS of the normalized light curve from XO-4. This process is repeated for each night to create the lowest RMS light curves possible and the final light curves are shown in Table 2.2.

Significant trends in the data are immediately apparent as the result of various red noise sources (Pont et al. 2006). These trends were mostly due to issues with the mount’s ability to track and guide combined with imperfect flat fielding and fringe

2http://www.astro.louisville.edu/software/astroimagej/

19 corrections. As a result the trends were correlated with XO-4b’s x and y location on the chip that vary over the course of observations. After detrending the data, the 7 individual light curves are shown in Figure 2.1 with the binned data in Figure 2.2. A summary of the planetary parameters derived from the transits are shown in Table 2.3. Even after detrending the data, evidence of red noise is still visible in the residuals of the light curve.

Detrend Parameters

EXOFAST can take trends in the photometric data and remove them to improve the quality of data. We investigate which, if any, detrend parameters in the DEMONEX data are significantly affecting the data quality and thus should be removed. We consider the the position on the star on the CCD, x and y, the time (BJDTDB), and airmass sec z as these are typical detrend parameters used in other studies (e.g., Collins et al. 2014). We take the case of no detrending and compare the results when the light curves were detrended against a single parameter. We then began adding in additional detrend parameters to detrend simultaneously. We notice that there are significant trends in all four detrend parameters when fitted individually. Once all of the parameters are included, the derived parameters converge to within 1 σ of values previously published in the literature as well as the values derived when we use only the N10 data in EXOFAST from §2.4.1. This suggests that including all available detrend parameters improves our analysis, at the expense of more computational time and creating the potential of a new local minima in the χ2 space that we search through. As airmass and time are correlated we only include time to reduce the parameter space that the EXOFAST MCMC chains must search through and to speed up the computational time. Although time is linear during each night and airmass is not, detrending against the two produce similar changes in

20 the derived parameters as seen in Figure 2.3. Thus our ﬁnal DEMONEX data set is detrended against x, y, and BJDTDB.

The parameter most sensitive to the quality of data and selection of detrend parameters is RP /R∗. Gillon et al. (2007) found a similar dependence with RP /R∗ when reducing their light curves using various aperture sizes. However many of the derived parameters are correlated with RP /R∗, parameters such as the spin-orbit alignment λ, which depend on both RP /R∗ and v sin I∗, are still subject to the choice of included detrend parameters. Again we notice that most parameters converge to previous estimates as more detrend parameters are included as seen in Figure 2.3.

It it worth noting that for the parameter RP /R∗ the change in the inferred value for RP /R∗ can vary by over 3 σ depending on the choice of, and number of, detrend parameters. Thus the systematic uncertainty due to the selection of detrend parameters is far greater than the statistical uncertainty quoted and derived from the data itself.

Additionally we look at the eﬀects of detrending on the inferred TTVs by enabling the time of the central transit to be a free parameter, as shown in Figure 2.4. We should note that not all of the possible combinations of detrend variables were able to be selected as the increased freedom of ﬂoating central transit times and the detrending parameters resulted in many chains failing to meet the convergence criteria. To counteract this a Gaussian prior was placed on the period to allow the chains to converge. This prior is only used when detrending DEMONEX data only. In some cases the inferred variations in the derived O-C is consistent within the uncertainty regardless of which detrending variable is selected. There are still cases where the O-C varies at the few σ level, indicating that the choice of detrend variable once again matters and the presence of systematic uncertainties that exist in addition to the quoted statistical uncertainties. It is still the case that even when

21 using detrending parameters, multiple nights have an O-C that imply a signiﬁcant TTV.

It is not clear to the authors which detrend parameters, or combinations of detrend parameters, are the “correct” parameters to detrend against. Including detrend parameters has signiﬁcant implications on the derived parameters, especially

RP /R∗, but we do not yet have an objective way to determine which set of parameters is “correct”. We consider some metric involving χ2 minimization, but note that this assumes uncorrelated data which we know not to be the case. We understand that our data contain systematics, and detrending is one such effort to reduce the effects of these systematics. We also note that including all of the available parameters produces inferred values consistent with previous measurements, but this can vary at the 3 σ level to the inferred value when detrend parameters are not used. We therefore conclude that detrended data inherently contain additional systematic uncertainties, typically not quoted, and detrending must be applied carefully in order to avoid biasing the inferred parameters. Additionally readers should be cautious when trusting the inferred values of these parameters in this and other papers. We have made efforts to quantify and reduce these systematic uncertainties, but do not yet have an ideal way in which to eliminate them altogether. Our study of these effects is however a step in the right direction.

2.4.3. Combined Data Set

Having demonstrated that the DEMONEX data (when properly detrended) are consistent with the results from previous analyses, we now combine the light curve data from the detrended DEMONEX data, MC08 XO Extended Team data (XOET), N10 data (FLWO), and Todorov et al. (2012) optical data (UDEM) to create our combined primary transit data set. Additionally we detrend each of the

22 non-DEMONEX data sets against time as we noticed trends in the XOET data. This final data set includes 24 primary transit light curves of XO-4b, covering 21 different nights, in 5 different bands. There are 7 transits taken from the 0.5-m DEMONEX telescope, 4 transits taken with the FLWO telescope from the N10 data set, 9 transits from the XO Extended Team MC08 data set, and 4 transits from the UDEM telescope in the Todorov et al. (2012) data set. These are shown binned by telescope in Figure 2.5 and binned together in Figure 2.6. Combined there are 17 full transits containing both ingress and egress with multiple transits covered in multiple wavelengths or by different telescopes. The remaining light curves include 7 partials with 4 containing only the ingress and 3 containing only the egress.

Data Diagnostics

To look at the overall quality of our data, we look at the RMS of the residuals of the combined light curves shown in the bottom inset of Figure 2.6. We find that the weighted RMS of the ∼6700 data points within 4 hours of the central transit to have a fractional RMS in the residuals of the normalized flux of 0.00221, or a factor of 3.4 smaller than the transit depth. To test the quality of our data we also verify that √ the RMS decreases as t when the data are binned. We provide an Allan variance plot, shown in Figure 2.7, where we increase the bin-size of the combined data set light curves and verify that binning the data at larger intervals does in fact decrease √ the weighted RMS as t as expected if the errors are uncorrelated. When the data are binned at 5 min intervals, as shown in Figure 2.6, the fractional RMS in the residuals of the normalized flux decreases to 0.000276, or a factor of 27 smaller than transit depth.

The most natural interpretation of the uncertainties we quote on the derived parameters assumes that the photometric uncertainties in the data are both

23 uncorrelated and Gaussian distributed. In Figure 2.8 we plot the distribution of the residuals in the light curves from all of the data sets normalized by their uncertainties. We ﬁnd that the distribution is not perfectly Gaussian, with a larger number of points with values of (O-C)/σ near zero than expected. This implies that our process of scaling the uncertainties by a constant factor is not entirely capturing the true nature of the systematic errors.

Mass-Radius

In addition we test the effects of using two methods to resolve the mass-radius degeneracy. The default relation for EXOFAST is based on the Torres relation, while the updated EXOFAST uses the Yale-Yonsei (YY2) isochrones. The Torres relation is based on Torres et al. (2010) who provide empirical estimates of stars with precise mass and radius measurements to derive simple polynomial functions of Teff , log g∗, and [Fe/H] that yield M∗ and R∗ with scatter within the relations of 6% and 3%, respectively. As an alternative, we also use isochrones based on Yi et al. (2001) and Demarque et al. (2004) which provide sets of isochrones over a wide range of metallicities and ages scaled to the solar mixture. The update to the EXOFAST code and implementation is described in Eastman et al. (2015) and combines the spectroscopic constraint on Teff and an additional penalty when Teff differs from the YY2 isochrones with an model uncertainty of 50 K.

We use both relations in our analysis and ﬁnd that the constraints are tighter for M∗ and R∗ when using the YY2 isochrones. In the cases of the mass and radius of the star (and therefore the planet) the constraints are tighter by 50-95%. We adopt the results from the YY2 over those derived from the Torres relations as they provide increased precision. The two models provide consistent results, providing

24 conﬁdence that the parameters we infer are accurate to within the precision with which we can measure them given the quality of the data.

Eccentricity

While other groups report that the eccentricity of the system is consistent with zero (McCullough et al. 2008; Narita et al. 2010; Todorov et al. 2012), we verify this by performing an additional run where we allow e cos ω∗and e sin ω∗to be free

+0.012 +0.0089 parameters. We ﬁnd that e cos ω∗ = 0.0026−0.0066 and e sin ω∗ = −0.0072−0.028 , +0.025 +110 such that the eccentricity is small e = 0.014−0.010 and ω∗ = −62−41 is largely unconstrained. These results are consistent to Todorov et al. (2012) who ﬁnd

|e cos ω∗| < 0.004 at the 3σ level. Therefore we furthermore adopt the constraint that e = 0.

Spin-Orbit λ

We simultaneously fit radial velocity data taken by MC08 and N10. The N10 radial velocity measurements also contain RM measurements taken during the transit and the RM data are fit as well. We separate the RM data to fit independent zero points and to scale the errors of the two data sets independently as we do not expect the night-to-night stellar variability and the stellar jitter to have the same magnitude and have the N10 RV and N10 RM data sets. The radial velocity and RM fits are shown in Figure 2.9 and Figure 2.10 respectively. Because we scale the individual data sets, we find that both models have a similar global fit with only a ∆χ2 = 0.6 between the two, with the Hirano model having the lower χ2 value. However if we look at only the RM data and we use the Hirano scaled error bars on both data sets,

2 we ﬁnd that there is a ∆χRM = 2.5, again with the Hirano model having the lower χ2 value.

25 Gaudi & Winn (2007) have shown that there is strong degeneracy between v sin I∗ and λ for systems with central transits, i.e. low impact parameters. In this case, placing a Gaussian prior on v sin I∗ and miss-estimating ∆vRM will lead to a miss-estimation of λ. This is likely the case with XO-4b with an impact

+0.077 parameter of b = 0.230−0.078. To constrain λ we apply a Gaussian prior on v sin I∗ = 8800 ± 500 [m/s] taken from MC08. However, the OTS model requires a

+430 +9.0 higher v sin I∗OTS = 100−420 and lower λOTS = −20.6−8.0 to ﬁt the data. The Hirano +460 model is closer to the stellar prior with v sin I∗Hirano = 8680−440 and prefers a higher +8.8 spin-orbit misalignment of λHirano = −40.0−7.5. Figure 2.11 illustrates the v sin I∗-λ degeneracy and where the two models lie in this parameter space. We note that, although the two models agree at nearly 1σ, the implications are very diﬀerent. In one case, one could infer that the system is almost consistent with being aligned

(λOTS is consistent with zero at ∼ 2σ), whereas this is much less likely for the other model. So, the difference due to the choice of models is not simply quantitative, but is also qualitative. Given the effective temperature of the host star of Teff ∼ 6400 K, the inference that XO-4b’s orbit is misaligned with its host star would be consistent with the observed trend that Hot Jupiters orbiting hot (Teff > 6250 K) stars tend to have high obliquities (Winn et al. 2010; Albrecht et al. 2012). There are cases where the two methods are consistent (Winn et al. 2008; Johnson et al. 2008), and these are typically systems with large impact parameters i.e. b > 0.6.

Ultimately, we believe the Hirano et al. (2010) model and inferences to be more reliable given the low-impact parameter of XO-4b. This poses a potential complication for future modeling. While the updated Ohta et al. (2009) models can be generally applied to RM measurements, the Hirano et al. (2010) model requires estimating the model parameters p and q, which in turn requires performing the cross-correlation analysis and ultimately having access to the RV data itself or relying

26 on other groups to perform this analysis and publish their results. Without access to the RV data or to the derived values of p and q, one must adopt a general model, such as the Ohta model, which may lead to biased and/or incorrect inferences.

Transit Timing Variations

We also fit for the mid-transit times of each observed transit, in order to investigate the presence of TTVs. During this we are also able to refine the orbital period by fitting the observed mid-transit times shown in Table 2.6 with a linear function

Tc(E) = Tc(0) + EP, (2.6) where E is the Epoch.

For the night of UT 2007-11-03 (XOET), the best ﬁt mid-transit time results in an O-C of ∼ 2000 s. Upon further investigation, this outlier is likely the result of the noisy photometry from the beginning of the night. UT 2007-11-03 contains no clear ingress, and a feature that could be the egress. After detrending the light curve against time, as no other detrend variables are available, the ∼ 2000 s variation is preferred by EXOFAST when allowing for TTVs. In Figure 2.12 we show the light curve for the TTV and no TTV cases. When we allow the central transit time to vary, we ﬁnd that the RMS of the residuals decreases from 0.0047 to 0.0044, and that the χ2 decreases from 404 to 349.

The data suggest that the 2000 s TTV solution is the better ﬁt, but a visual inspection of Figure 2.12 and one could easily choose the no-TTV solution by eye. We do however have a strong prior against 2000 s transit timing variations and do not believe that the variation is astrophysical in nature, and is perhaps due to a false minimum in the χ2 ﬁt. Little statistical power is contained any single light

27 curve, and to improve the stellar and planetary constraints we omit this night as to not bias the reﬁned transit ephemeris. Omitting this night gives a reﬁned transit ephemeris for XO-4b of

P = 4.1250687 ± 0.0000024 d (2.7)

TC (0) = 2454758.18978 ± 0.00024 [BJDTDB] (2.8) with a reduced ﬁt of χ2/dof = 3.89. These values supersede the ephemeris values constrained by the RV data alone during the global ﬁt and appear in Table 2.5 while the RV data appear as footnotes.

There are still many nights where the observed mid-transit time is consistent with the presence of TTVs as seen in Figure 2.13. As shown from our detrending analysis it is possible that there are still systematic errors, in addition to our statistical errors, that we have not yet accounted for that could explain the nights that are still consistent with the presence of TTVs. We also note that these are present in all four data sets, not just DEMONEX data.

2.5. Results

In a series of tables we present the median values of the MCMC chains and their 68% intervals for the stellar parameters, planetary parameters, radial velocity parameters, primary transit parameters, limb-darkening parameters, and secondary eclipse parameters in Table 2.5. The ephemeris in Table 2.5 is constrained by the RV data alone, but we do include the reﬁned ephemeris from the mid-transit times as footnotes.

Using the YY2 mass-radius models and the Hirano based RM models we ﬁnd

+0.030 +0.042 a reﬁned stellar mass of 1.293−0.029 M and stellar radius of 1.554−0.030 R with

28 +0.10 +0.040 a companion planet mass of 1.615−0.099 MJ and planet radius of 1.317−0.029 RJ.

Additionally we find a refined ephemeris of T0 = 2454758.18978±0.00024 BJDTDBand P = 4.1250687 ± 0.0000024 days when fitting for the new ephemeris with the mid-transits as free parameters. We are able to confirm the spin-orbit misalignment

+8.8 of λ = −40.0−7.5 when using the Hirano based RM models.

In general we find an improved or comparable statistical precision for all available parameters. We note that the improved precision on RP /R∗ and other correlated parameters (e.g. i and a/R∗) are subject to additional systematic uncertainties. There is significant disagreement between the inferred period (20σ) compared to the results published by MC08 and a 6σ difference between the inferred period and the N10 published value. This is similar to the 16σ disagreement between the N10 and MC08 periods.

2.6. Chapter Summary

We provide 7 new nights of observations of the misaligned Hot Jupiter XO-4b from the DEMONEX telescope and combine that data with previously released data to produce refined stellar and planetary parameters for the XO-4 system analyzed in a homogeneous manner. We investigate a number of possible combinations of detrend parameters and find that the quality of DEMONEX data is significantly improved when detrended against XO-4’s position on the CCD and against time. We also note that there is a 3σ difference in the inferred value of the parameter RP /R∗ depending on the choice of detrend parameters, and as such, caution should be exercised when detrending data as to not bias results with incorrect inferences.

After investigating both the Torres relation and the Yale-Yonsei isochrones, we are not yet sensitive to the choice of method to resolve the mass-radius degeneracy.

29 We are sensitive to the choice of Rossiter-McLaughlin model used to infer the projected stellar spin-planet orbit angle λ. We believe the Hirano et al. (2010) model and inferences to be more reliable; however, the Ohta et al. (2009) models can be generally applied to RM measurements. The Hirano et al. (2010) requires having access to the radial velocity data, or a previous estimate of p and q from the cross-correlation analysis. Without such access, one must adopt a general model, such as that based on Ohta et al. (2009), which we have shown may lead to biased and/or incorrect inferences.

30 Fig. 2.1.— Each of the 7 DEMONEX light curves with an arbitrary offset fitted using only DEMONEX data and detrended against the x and y position of the star on the chip, and the time (BJDTDB). The error bars are plotted in gray and the best-fit model is over plotted in red.

31 Fig. 2.2.— The binned data and the best-ﬁt model are shown at the bottom with residuals. Binned data is not used in the analysis but shown to better display the overall quality of the data and the statistical power of the DEMONEX data.

32 Fig. 2.3.— Various derived parameters in the DEMONEX light curves as a function of the chosen detrend parameters: the position of the star on the CCD (x, y), time t, and airmass sec z. The x-axis containing the detrend parameters is the same for each plot including detrending against multiple parameters simultaneously. The solid and dashed lined represent the published Narita results and their error bars. In most cases the inferred value is still consistent, i.e. within 1 σ, with previous results when no detrending parameters are chosen, but the answers tend to converge to more consistent values when a larger number of detrend parameters are included. It should be noted that for the parameter RP /R∗ the change in the detrend parameter can change the inferred value by over 3 σ and the systematic uncertainty due to the selection of detrend parameters is far greater than the statistical uncertainty estimated from the data itself. We did not apply an additional Gaussian prior on Tc (see §2.4.1) which explains the diﬀerence in our inferred value for the period relative to the published Narita value.

33 Fig. 2.4.— Various derived oﬀsets from the calculated central transit time to the measured Tc as a function of the various combinations of detrending parameters used for the 7 DEMONEX nights. For some nights the changes in O-C are consistent within the errors, while for others the inferred O-C is dependent on the choice of detrend parameters. It should be noted that most of the inferred O-C values are still inconsistent with zero and suggest the presence of a transit timing variation independent of the choice of detrend parameter. We believe that these TTVs are unlikely to be real, but rather are the result of as-yet unrecognized systematics.

34 Fig. 2.5.— Binned light curves from each telescope to show relative quality of data from each data source. The best-ﬁt global model is plotted in red. Each set of data is binned in the same way. The DEMONEX data is detrended against x, y, and t while all others are detrended against t. Binned data is not used in the analysis but shown to better display the overall quality of the data.

35 Fig. 2.6.— All of the data globally fit by EXOFAST from the four different telescopes (XOET having 9 nights, FLWO 4 nights, UDEM 4 nights, and DEMONEX having 7 nights), phased and binned. Over-plotted in red is the best fit model for the global analysis and the residuals of the binned data are shown at the bottom. Binned data is not used in the analysis but shown to better display the overall quality of the data and the statistical power of the total data set.

36 Fig. 2.7.— The RMS of the residuals of the individual data points from all of the light curves at various bin sizes in time. The weighted RMS of the unbinned data is √0.00221 as seen in the data point to the top left. We ﬁnd that the RMS decreases as t as expected in the photon-noise limited regime if the data points are uncorrelated. This decreases to 0.000276 when binned on 5 min intervals as shown in Figure 2.6.

37 Fig. 2.8.— The distribution of the residuals of the individual data points from all of the light curves normalized by their uncertainties. We note that we have scaled the uncertainties in the individual light curves by a constant multiplicative factor in order to force the χ2/dof ∼ 1. We ﬁnd that the distribution is not perfectly Gaussian, with a larger number of points with values of (O-C)/σ near zero than expected.

38 Fig. 2.9.— Phased radial velocity curves for the two data sets used. The best ﬁt model is over plotted in red with residuals below. The N10 data set is split into data during transit (N10 RM) and data out of transit (N10 RV). This is done to get a more accurate estimate of the error scaling on the Rossiter-McLaughlin data.

39 Fig. 2.10.— The N10 Rossiter-McLaughlin data set and the two best fit models. Residuals relative to the Hirano model are shown below. The two models fit the data 2 with two different values of λ that vary at the 1-σ level. There is a ∆χRM = 2.5 between the two RM data sets using the two RM models with the Hirano model having the lower χ2 value.

40 Fig. 2.11.— 68, 95, and 99% contours for the v sin I∗ and λ for the two Rossiter- McLaughlin models derived using the YY2 isochrones. These are equivalent to 1, 2, and 3σ error contours. Over-plotted is the stellar prior on v sin I∗ from McCullough et al. (2008). The two models prefer values of λ that diﬀer at the 1σ level. The Ohta model is consistent with λ = 0 at the 2σ level.

41 TTV,RMS=0.0044,

noTTV,RMS=0.0047,

Fig. 2.12.— The two light curves for UT 2007-11-03. In blue is the ﬁt where the central transit time is allowed to vary, while the bottom black curve has all of the central transit times ﬁxed to a linear ephemeris. The top curve has a lower residual RMS and χ2, but inferred a ∼ 2000 s transit timing variation.

42 Fig. 2.13.— O-C diagram for the calculated versus measured central transit times, the epochs are taken relative to the RV data. Nights with zero transit timing variations will lie on the dashed line. Even with the exception of UT 2007-11-03 (Epoch -74), there are still a few significant outliers and nights consistent with the presence of transit timing variation. The ∼ 2000 s outlier is the result of poor photometry and the fitting procedure and is omitted in our analysis. The refined transit ephemeris is T0 = 2454758.18978 ± 0.00024 [BJDTDB] and P = 4.1250687 ± 0.0000024 d.

43 √ 2 2a Parameter EXOFASTMCMC EXOFASTχ2 Naritaχ2 |∆|/ σ + σ

P [days] 4.1250801 ± 0.0000028 4.1250801 ± 0.0000028 4.1250828 ± 0.0000040 0.55

TC [BJDTDB] 2454485.93346 ± 0.00023 2454485.93346 ± 0.00023 2454485.93323 ± 0.00039 0.51 K [m/s] 172 ± 11 174 ± 11 168.6 ± 6.2 0.43 +470 v sin I∗ [m/s] 8660−450 8820 ± 530 8900 ± 500 0.11 +9.2 +8.1 λ [degrees] −41.2−7.5 −47.0 ± 14 −46.7−6.1 0.02 +0.12 a/R∗ 7.58−0.19 7.73 ± 0.23 7.68 ± 0.11 0.20 44 +0.00055 RP /R∗ 0.08770−0.00052 0.08739 ± 0.00060 0.0881 ± 0.0007 0.77 +0.60 i [degrees] 88.30−0.73 89.01 ± 0.87 88.8 ± 0.6 0.20 u2 0.3021 ± 0.0022 0.3098 ± 0.0027 0.35 ± 0.11 0.37 γ1 [m/s] −3.3 ± 8.1 −4.3 ± 8.4 −0.1 ± 2.9 0.47

a p 2 2 Deﬁned as |EXOFASTχ2 − Naritaχ2 |/ σEXOFAST + σNarita

Table 2.1. XO-4b values compared to the literature. Time[BJDTDB] Normalized Flux Flux Error x[pixels] y[pixels]

20081111: 2454782.762028 1.001644 0.002393 866.200393 1219.753505 2454782.762820 1.008380 0.002104 870.227406 1220.639406 2454782.763614 1.016223 0.001979 873.242392 1220.334975 2454782.764407 1.004863 0.001808 875.858008 1220.666278 2454782.765201 1.005626 0.001758 877.882314 1219.812195 2454782.765994 1.005393 0.001807 879.948887 1218.680215 2454782.766786 1.010737 0.001784 881.385388 1218.341993 2454782.767579 1.006882 0.001733 881.728714 1217.411449 2454782.768372 1.002178 0.001764 883.499507 1216.778263 2454782.769167 1.007074 0.001747 883.487292 1216.278131

bThis table is available in its entirety in the online journal. A portion is shown here for guidance regarding its form and content.

Table 2.2. DEMONEX sample data including trend parameters for XO-4b.b

45 Parameter [Units] Value

Primary Transit Parameters:

RP /R∗ 0.0893 ± 0.0012 +0.11 a/R∗ 7.70−0.21 +0.58 i[degrees] 88.47−0.81 +0.10 b 0.205−0.076 +0.00022 δ 0.00797−0.00021

TFWHM [days] 0.1672 ± 0.0011 +0.00096 τ[days] 0.01569−0.00050 +0.0016 T14[days] 0.1831−0.0014 +0.0033 PT 0.1183−0.0017 +0.0040 PT,G 0.1414−0.0021

u1Sloanz 0.1745 ± 0.0059 +0.0021 u2Sloanz 0.3019−0.0022

Table 2.3. Transit Parameters of XO-4b with DEMONEX.

46 O−C c Epoch TC Error O-C Error Group (BJDTDB) (s) (s)

-77 2454407.535746 266 -1971.90 -7.40 X -76 2454411.686963 268 287.14 1.07 X -74 2454419.936177 100 207.01 2.06 X -61 2454473.554962 140 -409.37 -2.92 X -60 2454477.685305 105 46.15 0.44 X -58 2454485.932467 175 -211.27 -1.21 X -58 2454485.938725 201 329.42 1.64 X -58 2454485.933786 176 -97.31 -0.55 X -53 2454506.560261 77 -0.41 -0.01 X 5 2454745.815586 59 105.39 1.77 F 6 2454749.939938 63 43.29 0.69 F 14 2454782.941958 102 168.95 1.65 D 21 2454811.815800 97 26.14 0.27 U 28 2454840.692756 104 152.38 1.45 U 29 2454844.814164 95 -164.08 -1.72 U 30 2454848.938339 240 -241.46 -1.00 D 36 2454873.688662 102 -250.21 -2.44 D 86 2455079.947141 91 176.96 1.94 F 102 2455145.946277 89 4.57 0.05 D 108 2455170.692531 129 -355.73 -2.75 D 124 2455236.695540 63 -193.49 -3.04 D 124 2455236.698073 32 25.36 0.78 F 133 2455273.825430 102 174.03 1.69 D 293 2455933.834816 79 7.64 0.10 U

cX=XOET, F=FLWO, D=DEMONEX, U=UDEM

Table 2.4. Transit Times for XO-4b

47 Parameter Units Value

Stellar Parameters: +0.030 M∗ Mass ( M ) 1.293−0.029 +0.042 R∗ Radius ( R ) 1.554−0.030 +0.28 L∗ Luminosity ( L ) 3.63−0.24 +0.023 ρ∗ Density (cgs) 0.486−0.031 +0.013 log g∗ Surface gravity (cgs) 4.166−0.018 +69 Teff Effective temperature (K) 6390−70 [Fe/H] Metallicity −0.040 ± 0.030 +460 v sin I∗ Rotational velocity (m/s) 8680−440 +8.8 λ Spin-orbit alignment (degrees) −40.0−7.5 Planetary Parameters: e Eccentricity 0 (assumed) P Period (days) 4.1250687 ± 0.0000024d a Semi-major axis (AU) 0.05485 ± 0.00042 +0.10 MP Mass ( MJ) 1.615−0.099 +0.040 RP Radius ( RJ) 1.317−0.029 +0.081 ρP Density (cgs) 0.873−0.084 +0.032 log gP Surface gravity 3.361−0.034 +25 Teq Equilibrium temperature (K) 1641−23 +0.0070 Θ Safronov number 0.1036−0.0068 9 −1 −2 +0.10 hF i Incident flux (10 erg s cm ) 1.646−0.089

(cont’d) Table 2.5. Median values and 68% conﬁdence interval for XO-4b.

48 Table 2.5—Continued

Parameter Units Value

RV Parameters: e TC Time of inferior conjunction (BJDTDB) 2454758.18978 ± 0.00024 K RV semi-amplitude (m/s) 172 ± 10. +3.5 KR RM amplitude (m/s) 66.4−3.4 +0.10 MP sin i Minimum mass ( MJ) 1.615−0.099 +0.000073 MP /M∗ Mass ratio 0.001192−0.000071 +0.0071 u1 RM linear limb darkening 0.3601−0.0069

u2 RM quadratic limb darkening 0.3097 ± 0.0026 +8.4 γ1 Systemic velocity (m/s) −4.3−8.5 +2.8 γ2 Systemic velocity (m/s) 0.2−2.7

γ3 Systemic velocity (m/s) −1 ± 16 +0.00000044 f(m1, m2) Mass function ( MJ) 0.00000229−0.00000038

(cont’d)

49 Table 2.5—Continued

Parameter Units Value

Primary Transit Parameters: +0.00050 RP /R∗ Radius of the planet in stellar radii 0.08712−0.00048 +0.12 a/R∗ Semi-major axis in stellar radii 7.59−0.17 +0.61 i Inclination (degrees) 88.26−0.63 +0.077 b Impact parameter 0.230−0.078 +0.000087 δ Transit depth 0.007589−0.000083

TFWHM FWHM duration (days) 0.16880 ± 0.00051 +0.00078 τ Ingress/egress duration (days) 0.01562−0.00053 +0.00092 T14 Total duration (days) 0.18448−0.00079 +0.0027 PT A priori non-grazing transit probability 0.1203−0.0018 +0.0033 PT,G A priori transit probability 0.1432−0.0022

u1B Linear Limb-darkening 0.514 ± 0.012 +0.0079 u2B Quadratic Limb-darkening 0.2544−0.0080 +0.0063 u1I Linear Limb-darkening 0.2064−0.0062

u2I Quadratic Limb-darkening 0.3091 ± 0.0021 +0.0067 u1R Linear Limb-darkening 0.2762−0.0066

u2R Quadratic Limb-darkening 0.3200 ± 0.0022 +0.0060 u1Sloanz Linear Limb-darkening 0.1744−0.0059 +0.0021 u2Sloanz Quadratic Limb-darkening 0.3020−0.0022 +0.0071 u1V Linear Limb-darkening 0.3601−0.0069

u2V Quadratic Limb-darkening 0.3097 ± 0.0026

Secondary Eclipse Parameters: +0.083 TS Time of eclipse (BJDTDB) 2454483.931−0.11

d +0.00061 From RV only: P = 4.12473−0.00047 days

e +0.082 From RV only: TC = 2454485.993−0.11 BJDTDB

50 Chapter 3: DEMONEXT

This chapter is based on work published in Villanueva et al. (2018a).

3.1. Introduction

Time-series photometry is a useful tool for studies of exoplanets, supernovae, eclipsing binaries, and active galactic nuclei. Much of the night sky is dynamically changing, and synoptic observations can provide essential information. Despite the ever increasing demand for larger-aperture telescopes on the ground, as well as new and larger-aperture space-based telescopes, ground-based time-series photometry from small, half-meter aperture telescopes continues to be a powerful tool in modern astronomy. Many surveys continue to rely on small ground-based observatories as the workhorses of the follow-up observations and including facilities such as Las Cumbres Observatory (Brown et al. 2013) and Minerva (Swift et al. 2015). In addition to their use to study known variable objects, such telescopes can play an important role as survey instruments, as demonstrated by a number of recent discoveries by surveys such as MEarth (Nutzman & Charbonneau 2008) and Trappist (Gillon et al. 2016).

With the large number of targets to be discovered by upcoming all-sky surveys like TESS (Ricker et al. 2015; Sullivan et al. 2015) and LSST (Ivezic et al. 2008), the demand for additional observations from these half-meter telescopes is unlikely to decrease in the near future. Due to the expected large numbers of transients from these surveys, a more eﬃcient and eﬀective follow-up tool is required, and

51 suggests the need to develop and deploy more fully robotic telescopes. With this in mind, The Ohio State University and Vanderbilt University have come together to develop a new robotic observatory called the DEdicated MONitor of EXoplanets and Transients (DEMONEXT).

This paper is outlined as follows, in Section 3.2 we describe the scientiﬁc motivation for building the telescope. The optical and mechanical system is described in Section 3.3. Nightly operations are described in Section 3.4. The data products and system performance are described in Sections 3.5 and 3.6. A summary of ﬁrst year science results can be found in Section 3.7, and conclusions in Section 3.8.

3.2. Science Drivers

In order to maximize both the quantity and quality of data from DEMONEXT, we must ﬁrst consider the science drivers for developing and deploying a relatively small-aperture, robotic telescope. There are two ongoing surveys that both The Ohio State University and Vanderbilt University are invested in, and what follows are the stated goals from the joint Ohio State and Vanderbilt community for DEMONEXT. These goals dictate the hardware selection, software design, and the operations model we have adopted.

3.2.1. KELT Exoplanet Follow-Up

The Kilodegree Extremely Little Telescope (KELT) ground-based transit survey (Pepper et al. 2007; Siverd et al. 2012; Pepper et al. 2012; Kuhn et al. 2016) has reported sixteen exoplanet discoveries thus far1, including one joint discovery (Temple et al. 2017), and one simultaneous discovery (Lund et al. 2017; Talens

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

52 et al. 2017), and has more planets in preparation (KELT Collaboration, private communication). KELT is primarily sensitive to hot Jupiters around bright host stars, typically V < 13. KELT candidates’ primary transits occur once per cycle with periods from days to weeks with the transit duration lasting 1–8 hours. Although the fraction of time for which one would want to make observations for any individual system is low, with close to 300 objects currently on the KELT North and South candidate lists, there is no shortage of candidates requiring observations each night. To constrain the planetary nature and to aide in future scheduling, it is crucial to observe the ingress or egress of each transit, as this conﬁrms the depth and timing of future transits. Scheduling observations around hundreds of candidates only during transit becomes critical. Primary transit events for these massive planets have depths of 1% or less, with millimagnitude (mmag) precision photometry required over the full duration of the event. Using relative aperture photometry, this can be achieved with 10–30 comparison stars in a single ﬁeld-of view (FOV). During the transit event observations should be taken at the highest cadence possible, with exposure times of seconds to a few minutes, to constrain the shape, parameters, and timing of individual transit light curves.

To make these observations we require a telescope with sufficient aperture to make mmag precision follow-up observations on V < 13 targets. This can be achieved with < 1 meter telescopes. However, to avoid saturation, the telescope must be defocused prior to observations. To obtain a sufficient number of comparison stars using relative aperture photometry we require a ∼ 0.5 degree diameter FOV. The telescope and mount should be stable, i.e. should achieve sub-pixel pointing, over the full duration of the observation, which may last over eight hours. This also requires continuous, uninterrupted observations, which excludes mounts that require meridian flips. The telescope must also have the ability to schedule around all of the

53 300 targets in a way that ensures that either an ingress or egress, preferably both, are observed for each transit, along with some amount of both out-of-transit and in-transit data.

3.2.2. ASAS-SN Transient Follow-Up

The All-Sky Automated Search for SuperNovae (ASAS-SN)2 surveys the entire visible sky every night down to V = 17 (Shappee et al. 2014). Because ASAS-SN discovery images have large pixels, transient candidates require follow-up observations with lager telescopes and finer sampled images. Follow-up observations require a prompt single-epoch V -band image for candidate confirmation. It is important to confirm the transient nature as quickly as possible, in order to begin characterization as early as possible. Following confirmation, a month-long observing campaign is initiated to characterize the photometric evolution of the transient at the 1-10% level. This campaign consists of multi-epoch, multi-band observations at 1–4 day cadence.

Follow-up of ASAS-SN targets requires a facility capable of both 1 − 10% photometry in multiple bands at daily to weekly cadence, and a tool for prompt observations. For the faint (V = 17) targets, 1 − 10% photometry can be achieved in a single 5 minute exposure with a half-meter telescope. The guiding requirements are less strenuous. The pointing is required to drift less than a pixel over the exposure to ensure images are not elongated due to tracking errors. A filter wheel with a full compliment of either Sloan g0, r0, i0, and z0 or Johnson-Cousins B, V , R, and I filters are required for multi-band characterization. The scheduler must be able to observe objects in a flexible manner, and ASAS-SN observations are a natural fit to fill in

2http://www.astronomy.ohio-state.edu/∼assassin/index.shtml

54 gaps between KELT observations. Additionally, prompt follow-up of recent events is preferred for targets that are identiﬁed in real time.

3.2.3. Ancillary Science

Given the strict scheduling requirements of KELT follow-up observations, and the slow-cadence required of ASAS-SN follow-up observations, there is a significant amount of free time left in the DEMONEXT queue to fill. With observations scheduled around high-cadence KELT or long-term monitoring of ASAS-SN targets, any science that can be done with either long or short cadence time-series photometry around V < 20 sources are ideal targets for DEMONEXT. Within those parameters, there have been no shortage of suggestions from the DEMONEXT community on how to fill this time. Some of the more popular uses have been Galactic microlensing follow-up, monitoring of Active Galactic Nuclei (AGN), studies of stellar variability, stellar rotation, and eclipsing binaries. Many of these applications are described in Section 3.7.

3.2.4. Summary of Science Requirements

In summary, DEMONEXT is required to make observations in two modes: a high-cadence, short-exposure time mode, around bright V < 13 targets, and a low-cadence, long-exposure time mode around fainter V < 17 targets. The telescope is required to automatically focus and defocus, depending on the target. For the low-cadence mode, guiding need only be stable enough to keep images from becoming elongated. However, for the high-cadence mode, in order to achieve ∼mmag photometry, sub-pixel stability is required over many hours with uninterrupted observations. The telescope must also be able to schedule around

55 ingresses and egresses of roughly 300 KELT candidates, while filling in the gaps with observations of non-KELT targets. We also require a full set of either Sloan or Johnson-Cousins filters, with the ability to take multi-band observations. Finally, we need tools for prompt follow-up of rapidly-identified transient targets.

3.3. Optical and Mechanical System

The optical and mechanical details for DEMONEXT are largely unchanged from those reported in Villanueva et al. (2016a) at the time of commissioning. We present a summary and description of the optical and mechanical system here, including updates made since commissioning in May 2016. A photograph of the ﬁnal operational system is shown in Figure 3.1, and a summary of the primary components is given in Table 3.1. All components were purchased from vendors and integrated into one system by the authors. The three dominate costs were the telescope, mount, and science camera (30,000-45,000 USD each), and the entire system could be replicated for under 200,000 USD. Only a small number of parts, like mounting adapters or custom cables, were fabricated in-house.

3.3.1. Optical Telescope Assembly and Mount

The DEMONEXT optical telescope assembly (OTA) is a PlaneWave Instruments 20 inch (0.5 meter) CDK20 telescope. The oﬀ-axis performance of the CDK20 allows for a ﬁeld-of-view (FOV) greater than half-a-degree depending on the size of the detector array. The CDK20 telescope is mounted on a Mathis Instruments MI-750/1000 equatorial fork mount. This design allows for continuous and uninterrupted observations through the meridian, required for observations of long duration transit events or monitoring programs.

56 3.3.2. Camera, Filter, and Focuser

The science CCD and filters were re-purposed from DEMONEXT’s predecessor, the DEdicated MONitor of EXotransits (Eastman et al. 2010a). The imager is a 2k×2k Fairchild CCD3041 detector packaged by Finger Lakes Instrumentation (FLI) with a 0.90 arcsecond per pixel scale, and 30.7×30.7 arcminute FOV. The pixel scale is well matched to the typical seeing of ≈ 2 arcseconds at Winer Observatory. DEMONEXT users are discouraged from requesting observations of high density fields or those requiring high spatial resolution, although users have had some success as shown in Section 3.7.3. The large FOV results in an increased number of comparison stars available for differential photometry, the primary method of obtaining time-series photometry for the DEMONEXT community. DEMONEXT is equipped with a ten position FLI CFW-3-10 filter wheel that holds 8 filters: Sloan g0, r0, i0, z0, Bessell (Johnson/Cousins) B, V , R, I, and clear filters. We plan to add the final filter once a decision is made by the DEMONEXT user community. The CDK20 is equipped with a Hedrick electronic focuser, and PlaneWave;s electronic focus accessory (EFA) kit allows remote remote control of the telescope focus position with the PlaneWave Interface 3 (PWI3) software and drivers3.

3.3.3. Guide Scope

A SBIG ST402-ME guide camera and an Orion 80 mm ED refractor telescope were installed in October 2016 to auto-guide during continuous observations over many hours and for long (> 5 min) science exposures. The guide scope is aligned with the CDK20 and guides on the brightest star in the FOV. When used for KELT follow-up, the KELT target star is typically selected. For other observations, a guide

3http://planewave.com/downloads/software/

57 star is selected within the FOV with guide camera integration times of < 10 seconds. The guide camera is not equipped with an automated focuser, and was focused manually during the installation. The focus does not change enough over time to require focusing more than once per year after summer shutdown restart.

3.3.4. Remote Power and Connectivity

Remote power management is implemented using an IP Power 9258 four outlet power supply. A single AC power cable is run to the power supply, which is connected to the science camera, guide camera, and mount. The IP Power 9258 allows us to individually power cycle each device via a local network connection. Each accessory’s AC cable has an AC-to-DC converter. The main source of AC power for all telescope systems is ﬁltered through a Powerware 9125 uninterruptable power supply. The focuser, ﬁlter wheel, and guide camera are powered using the accessory hub on the science camera.

The focuser, filter wheel, and guide camera are USB controlled through the accessory hub on the science camera, and the science camera, guide camera, and mount are USB controlled as well. To both minimize the number of cables that have to be run from the telescope to the control computer and overcome the 5 meter limit of most passive USB cables, we use an Icron 2304 four USB-to-Ethernet converter to convert the three USB lines to a single Cat 5 cable. We then convert that Cat 5 cable and IP Power 9258 Ethernet cable to a single fiber optic cable pair to transmit the devices over the 12.5 meters to the computer using a FiberTronics ESW-605 Ethernet-to-fiber switch. The only cables that run the full distance between the warm room and the telescope are a weatherproof AC power cable and an armored fiber optic cable pair. We convert the fiber optic back to Cat 5, and the Cat 5 to USB using the receivers of both switches.

58 3.4. Nightly Operations

3.4.1. Software

The automation and scheduling software for the operational version of DEMONEXT are derived from those described in Villanueva et al. (2016a), but have been updated and are described in their current state here. A summary of the software and the components they interface with is given in Table 3.2. All devices and software are Astronomy Common Object Model (ASCOM)4 compatible, and come with a set of common scripted commands. An Optiplex 9020 desktop computer running Windows 7 has Python, MaximDL, PWI3, and Sidereal Technology software installed to execute all of the scheduling and software. All control programs were written by us in Python, using the Anaconda version of Python5 with the PyEphem6, Numpy7, astropy8, and CV29 modules. The Python programs written for DEMONEXT are based on older programs implemented in Visual Basic for the original DEMONEX system (Eastman et al. 2010a), with new programs written speciﬁcally for DEMONEXT, or adapted from programs developed for the MINERVA Project10 (Swift et al. 2015), which uses a similar mount to DEMONEXT. We also wrote a Python wrapper to initialize the auto-focus routine that comes with the EFA focus software. All of our Python code is available under the Creative Commons.

4http://ascom-standards.org/ 5https://www.continuum.io/anaconda 6http://rhodesmill.org/pyephem/ 7http://www.numpy.org/ 8http://www.astropy.org/ 9http://opencv.org/ 10https://github.com/MinervaCollaboration/minerva-control

59 3.4.2. System Initialization

During the afternoon of a scheduled night of operations, a master control program is executed to initialize the system. This program creates new engineering and observing logs for the night and performs various housekeeping tasks. This includes a query of the operations status ﬁle created at the end of each night to determine if DEMONEXT needs to execute hardware or software restarts to recover from fault conditions such as improper shutdown, power failure, etc. The night’s transit and continuous monitoring programs are then scheduled as described in Section 3.4.5.

3.4.3. Nightly Calibrations

Each night we acquire calibration images before starting science observing. Bias and dark images are acquired before the roof is opened. Sky flats are acquired after the roof is opened in the evening when the Sun is < 3 degrees below the horizon. Because the time window for sky flats is so narrow, instead of attempting attempt to take flats in multiple filters during either morning or evening twilight, we take sky flats in the filter with the longest time elapsed since the last sky flat for that filter was taken. We scale the exposure times following each exposure to maintain counts in each sky exposure between 10–40k counts. On average, 14-17 flats with suitable counts are taken in two filters each night. This allows us to create a new master flat in each filter every 4–6 days depending on weather. Each master flat image has the standard deviation over the entire flat recorded along with the date. Standard deviations are 3–5% across the illumination pattern. The best master flat used in calibrations each night is the master flat from all previous master flats with the lowest standard deviation. However, we add a constant term (0.01%) to the standard deviation for each night elapsed since the observation to keep flats current.

60 3.4.4. Observations

The carbon fiber truss structure of the CDK20 makes for only small variations in focus with temperature. We focus the telescope twice per night: before the first science observations, and at the first convenient break between programs after midnight. We find the best focus in the V band filter to within ∼ 100 µm. Focus offsets relative to the V band were measured in all other filters and the offsets range from 60-140 µm. We note that these are all consistent with zero offset when considering the measured uncertainty for the best position each (∼ 100 µm). Nonetheless, we use the offset in each filter to re-position the focuser when changing the filter during observations.

Science programs begin execution when the Sun is > 18 degrees below the horizon, and continue until either the Sun is < 18 degrees below the horizon or no science targets remain in the queue that can be completed before the end of astronomical twilight. DEMONEXT’s operations are divided into two observing modes: objects requiring continuous observations (e.g. KELT Transit Targets) and the long-term monitoring queue (e.g. ASAS-SN Monitoring Programs), which require observations over many epochs. We do not yet utilize a mode for rapid follow-up as discussed in Section 3.7.2.

3.4.5. Continuous Time-Series Observations

We define our continuous time-series observations as a set of images taken in a continuous series, with the cadence set by the exposure time and overheads between exposures. We utilize this mode to take a series of a hundred to a thousand images consecutively on a single target over a span of one to eight hours. This can be done in a single filter or in an alternating sequence of filters. We use this mode for all of

61 our observations of exoplanet transits from KELT or other target lists, monitoring of high priority Galactic micro-lensing events, and studies of stellar rotation.

DEMONEXT queries a list transiting exoplanet candidates, other targets requiring continuous observations, and a proprietary list of planetary candidates maintained by KELT known as the K-list. DEMONEXT calculates which targets’ ephemerides indicate a transit will occur during the upcoming night while the target is visible. Viable candidates are sorted by priority, and the Julian Date (JD) of the start and end times for observations are scheduled. Excluding the times of observations of scheduled targets, any new targets that meet all of the requirements and can also be scheduled are included. The same process is repeated for the K-list, with the highest priority KELT targets scheduled first, to fill in as much of the night as possible. A minimum of 90 minutes is required for KELT observations and we require either the full ingress or egress. We avoid filling the queue with only in- or out-of-transit events, as neither provide interesting constraints on the depths or refine the ephemerides. We average 0–3 KELT candidates scheduled if there are no other targets in the queue.

Throughout the night, DEMONEXT queries the current time and references the start and end times of the scheduled transits. If the current JD is in the window of the scheduled transit or continuous observation start and end times, DEMONEXT will begin continuous observations. DEMONEXT will begin by slewing to the target’s coordinates, moving to the best focus position(s) for the required filter(s), and sets the exposure time(s) if provided or calculates one according to a reference magnitude and a requested SNR. If multiple filters are requested, DEMONEXT alternates between each filter and sets the appropriate exposure time prior to each filter. Observations are repeated until the the scheduled end time. Including the 6.75 seconds of read time of the CCD, the total overhead between exposures is 12–18

62 seconds. This mode is used for covering transits or continuous-observation targets (e.g. observations of open clusters) for eight or more hours of uninterrupted imaging.

3.4.6. Long-Term Monitoring

We fill the times when no continuous observations are required by observing objects at low cadence in the long-term monitoring queue. The queue is populated by targets from a variety of sources, programs, and investigators and each target has a set priority, and a requested cadence. We use a dynamic scheduling system similar to the one used by the Remote Telescope System11 (Kubánek2008; Kubáneket al. 2012). Each object in the queue is given a score based on the current time, the time since the object was last observed, and the requested cadence of the observations. The score is modified by visibility constraints, priority, and partner time. The exact form of the scores are described in the Appendix of Villanueva et al. (2016a), although we modify these to meet our observational needs.

After calculating the score of each object in the queue, DEMONEXT observes the target with the highest score. We do not yet attempt to optimize the sequence of observations that would yield the highest set of scores, as there has yet to be any signiﬁcant complaints on the quantity or quality of data from the investigators. For each observation, DEMONEXT will slew to the target’s coordinates, re-position the focuser, set the ﬁlter wheel, calculate an exposure time, and acquire the requested exposures. The long-term monitoring queue is updated and the process is then repeated.

11http://rts2.org/

63 3.5. Data Processing

All data processing is initiated manually at OSU using IDL12 scripts each morning. Our IDL routines make use of the IDL Astronomy User’s Library13. Calibrated images are available on the OSU servers for OSU users around noon each day, with delays over weekends. Vanderbilt images are pushed to a server at Vanderbilt following the reductions where they become available to the investigators. All images taken each night, including science images as well as bias, dark, and sky ﬂats, are transferred to 10TB worth of storage at OSU, with a second 10TB hard drive for backup of all data. All data processing is done in IDL.

3.5.1. Bias, Dark, and Flat Corrections

Master bias and master dark frames are created by median combining all bias and dark images, respectively, which are used to bias and dark correct all images from that night. For the two filters of sky flats taken that night, a master sky flat is created by median combining the images in each filter. The master sky flats in every filter are stored in a separate directory. Each sky flat has the RMS recorded, and for each filter observed in that night, a single master sky flat in each filter is selected based on the RMS and time elapsed since the master sky flat was created. Each image is then flat corrected from the set of master sky flats chosen.

3.5.2. World Coordinate Solution

After bias, dark, and ﬂat correcting the images, we execute IDL scripts to use the plate scale of the telescope, target RA and DEC, and FOV as inputs that are fed

12harrisgeospatial.com/ProductsandTechnology/Software/IDL.aspx 13https://idlastro.gsfc.nasa.gov/

64 into SExtractor14 to extract the locations of the sources in each image. We use the output source locations and Astrometry.net15 to obtain World Coordinate Solution (WCS) on the calibrated images. The results are added to the ﬁts header for each image. On nights with poor weather there are often too few stars to obtain a solution. Images without WCS solutions are typically discarded during the later analysis as a ﬁrst order quality assurance check.

3.5.3. Reductions Left to Users

We provide users with science-calibrated images with WCS solutions. Given the number of DEMONEXT users and the variety of observations and science goals, we leave methods for analysis and time-series photometry up to individual DEMONEXT users. Each user is given the location on the OSU or Vanderbilt servers with their science calibrated images. Users are encouraged to copy the original images to their personal directories and perform their own aperture, PSF, or image subtractions routines.

3.6. System Performance

3.6.1. Noise

We investigate the noise performance of DEMONEXT photometry in the continuous time-series mode. The measured raw, unbinned, photometric noise as a function of magnitude is plotted in Figure 3.2 from 8 hours of continuous observations of the Alpha Perseus cluster in I. The dot-dash diagonal line represents the standard

14https://www.astromatic.net/software/sextractor 15https://www.http://astrometry.net/

65 signal-to-noise equation

S SNR = √ , (3.1) S + B + D + R2 where we include terms for the source flux S, background B, dark current D, and read noise R. We do not include terms for flat fielding errors, comparison stars noise, fringing or other systematic effects. We include a term for the scintillation noise N (Young 1967; Hartman et al. 2005)

−2/3 7/4 −h/8000 −1/2 N = N0d X e (2texp) , (3.2)

where N0 = 0.1, the diameter of the telescope d = 50 cm, and the altitude of the observatory h = 1515.7 m are fixed. The exposure time for these images is texp = 20 s and the airmass varied from X = 1.5–2.4 in these images. We plot a number of possible airmass terms in horizontal dashed lines. All lines, both SNR and scintillation, are calculated from the properties of the telescope and are not a fit to the data. We find that the data is well represented when the SNR and airmass X = 2.4 lines are combined. The gray line is the median of various bins in magnitude to the DEMONEXT data.

DEMONEXT regularly achieves 1% photometry for targets with sufficient counts (> 104) above the DEMONEXT airmass limit of X = 2.4. When near zenith, DEMONEXT should achieve mmag photometry on sufficiently bright stars (i.e. counts > 106), with 20 second exposures. In practice, DEMONEXT achieves 2–4 mmag photometry on V > 12 KELT targets over a range of airmasses and exposure times that correspond roughly to the scintillation noise. The scintillation noise floor is set by the photometric precision floor for most stars, but the bright end is dominated by a combination of scintillation noise, flat fielding errors, and fringing.

66 √ We also verify that the noise decreases as t as expected of white noise. This can be seen in Figure 3.3, where the raw, unbinned data of one of the brighter stars (V ≈ 11) is shown as the ﬁrst data point in the top left. The binned data and their √ uncertainties are the data points. The predicted white-noise slope of t is shown in the solid line, normalized to the RMS of the unbinned data point. The data follow this slope, with only marginal evidence of non-white noise at longer time-scales. We therefore expect mmag precision at 5–6 minute timescales.

An initial look at the distribution of the errors shows that the errors are Gaussian with a χ2/dof = 2.89 for a star with constant flux assuming photometric and scintillation noise. We model this data in Figure 3.4 shown as the black histogram, and the expected distribution is shown as the black dashed curve. The difference in the actual distribution of errors (solid histogram) and the expected distributions (curve) is likely due to the systematic errors from flat fielding or fringing uncertainties. We scale the errors in the data to achieve a χ2/dof = 1.0 as shown in the red histogram to account for the systematic uncertainties.

3.6.2. Guiding

DEMONEXT guiding errors can be described in two regimes: guiding errors incurred during a single exposure, and intra-exposure guiding errors that accumulate during continuous observations. For single exposures, guiding errors will cause individual point sources to be elongated. We have found that DEMONEXT drifts at a rate of up to ≤ 5 pix hr−1 (≤ 4.5 ” hr−1) when unguided. For a typical exposure, with a 5 minute integration time and a pixel scale of 0.9 ” pix−1, this causes the center of a point-source to drift by 0.083 pixels (0.075 ”). With a typical seeing of 2 ” or 2.22 pixels, this results in an upper limit for the elongation of a point source within a single exposure of ε ≤ 0.036 caused by guiding errors.

67 To obtain high quality, systematic-free light curves, DEMONEXT must minimize the effects flat-fielding errors, the loss of comparison stars that drift out of the FOV during observations, inter-pixel variations in sensitivity, and the effects of fringing (which DEMONEXT suffers from in the I and z0 bands), by keeping targets on the same pixel for observations as long as 8 hours or more. Prior to the installation of the guide scope, DEMONEXT was used without any guiding. With a drift rate of 5 pix hr−1, drifts of 40 pixels (36”) were found over the course of 8 hours. After the installation of the guide scope, the drift was reduced to to 3 pix hr−1 (2.7 ” hr−1). Drifts of 30 pixels (27”) were still found in the longest observations, consistent with differential flexure between the guide scope and the OTA. The camera is mounted on the back of the OTA (see Fig. 3.1), and a modification to the guide scope’s mount is planned for Fall 2017 to increase the rigidity and stability of the system. To meet our sub-pixel guiding requirement, the modified guide scope mount would have to reduce the drift rate to ≤ 0.1 pix hr−1 (≤ 0.09 ” hr−1), an order of magnitude lower than the current rate. We investigate alternative methods to overcome the mechanical limitation of the guide scope.

Science Guiding

To meet the sub-pixel guiding stability DEMONEXT switched to guiding on the science images during spring 2017. By guiding on the science images themselves, we bypass the guide scope and any errors due to differential flexure. Prior to twilight, DEMONEXT takes a series of five images. The first is a reference, followed by two images after moving the mount twice in only RA, ±α, and then two more images after moving twice only in DEC, ±δ. DEMONEXT then solves for the locations of the brightest 50 objects using routines in the Python OpenCV library. For the stars in the FOV of each image, we measure ∆x and ∆y in the image plane and then fit

68 for image rotation to map ∆α and ∆δ in RA and DEC, the relative rotation θ, and scale factors C1,C2:

∆α = C1∆x cos θ − C2∆y sin θ (3.3)

∆δ = C1∆x sin θ + C2∆y cos θ (3.4)

The orientation and scale do not change from night-to-night, but are required to be measured as the CCD is not perfectly aligned North and East. We ﬁnd that the CCD is mis-aligned with North up and East left to by ≈6 degrees.

During continuous observations (e.g. transits) the ﬁrst image is used as a reference image, with the locations of the 50 brightest sources recorded. Following each exposure (typically 20 seconds), DEMONEXT solves for the locations of the 50 brightest sources, and we solve for the ∆x and ∆y in the image plane to send a correction in RA and DEC (∆α,∆δ) to the mount prior to starting the next exposure. The total overhead is 12–18 seconds, and the additional overhead penalty due to science guiding is < 2 seconds, which is dominated by the time to execute the two slews. As the slew times are short < 1 second each, we are currently investigating ways to further reduce overheads.

We show two representative nights of science guiding in Figure 3.5. The black line represent a night with good weather and no clouds. There are N = 576 consecutive 20 second exposures (plus 12–18 seconds overhead), where target star stability is kept on the same pixel with the median deviation of 0.44 pixels and the peak in the distribution of 0.41 pixels. The red curve represents a night with poor observing conditions, with a constant layer of clouds. There are N = 443, 20 second images with a peak in the distribution of 0.61 pixels, but with a longer tail and median deviation of 0.95 pixels. We ﬁnd no evidence of long-term drift when using

69 science guiding over 5.5 and 4.4 hours of guiding respectively. We are able to keep the median deviation to < 1 pixel in poor observing conditions, and < 0.5 pixels in good observing conditions. This is a substantial improvement over either not guiding, or using the guide scope. We have not yet investigated the possibility of using both the guide scope and science guiding in parallel to further increase guiding stability. With science guiding, DEMONEXT now meets the sub-pixel guiding requirements set forth by our science drivers.

3.6.3. Automation Yields

One of the primary reasons for automating the DEMONEXT telescope was to increase the yield of observations of KELT transiting planet candidates. To date DEMONEXT has been operational for 291 nights and submitted 143 transit light curves, including multiple events on the same night, or the same event on one night but in alternating ﬁlters. This is a rate of 14.7 events per month, or 0.49 per night. Villanueva et al. (2016a) predicted a rate of 0.81 per night, but failed to include weather or ancillary science projects. Adjusting the Villanueva et al. (2016a) rate for weather gives a predicted rate of 0.65 per night, comparable to the rate 0.49 per night we achieved. Distributions of scheduled, observed, and submitted events from May 2016 through June 2017 can be seen in Figure 3.6.

Diﬀerences between the number of scheduled events (black) and observed events (blue) can be attributed to a combination of technical issues and weather. Diﬀerences between the number of observed events (blue) and those submitted (red) are due to the data taken being unusable. Winer observatory closes during the summer monsoon season and accounts for the lack of observations in June-September 2016. The lack of usable observations can be seen during the month of May 2016 while in the early days of commissioning and general debugging, during November

70 2016 through March 2017 when we had a run of poor weather resulting in occasional roof closures, and we were commissioning a number of new features during the ﬁrst year that all had problems to be solved. Small numbers of scheduled events can be seen during months cut short by the monsoon season (June 2016), during months with short nights (April–June 2017), and during months where other science projects were observed at a greater priority (December 2016). Cases where the data was unusable are usually due to weather, pointing or slew issues, and the failure of the original ﬁlter wheel (March 2017).

Overall DEMONEXT is performing well as a follow-up resource. We scheduled 301 transits in the first 291 nights of operation, averaging 31 events scheduled per month, excluding summer shutdown. Of those 301, 188 were observed (62%) and 143 were submitted (48%). Of those that were observed, 143 of 188 (76%) were of sufficient quality to submit to the collaboration. Over the first year of operation, DEMONEXT yields an average of 14–15 usable light curves per month. We expect this number could increase to 24 per month assuming 80% of nights are usable due to weather, zero mechanical or software malfunctions, and no other higher priority science programs.

3.7. DEMONEXT Science Results

3.7.1. KELT Exoplanet Follow-Up

DEMONEXT observed 188 planetary transit events around 116 unique targets and submitted 143 transit candidate events to KELT. Targets are transiting planet candidates, and DEMONEXT is used to vet candidates for astrophysical false positives e.g., nearby eclipsing binaries. KELT candidate host stars have magnitudes

71 that range from 7 < V < 13 with planets that produce transit depths of 1–10 mmag. Typical observations include a continuous series of exposures in a single band, most commonly i0 to minimize the effects of limb darkening. DEMONEXT also has the capability to observe in multiple filters to obtain wavelength-dependent transit depth measurements. For each target an exposure times is calculated to produce a SNR of ∼1000, which corresponds to exposure times of 120 seconds or less for most KELT stars. For objects that have exposure times less than 20 seconds, DEMONEXT adopts a minimum exposure time of 20 seconds and defocuses the telescope to spread the flux over more pixels to avoid non-linearity effects and saturation in the CCD. This is done to prevent the readout overhead from exceeding the shutter open time.

The DEMONEXT light curves for KELT-20b were featured in Lund et al. (2017), and we show these light curves in Figure 3.7. We note that this planet was independently discovered as MASCARA-2b (Talens et al. 2017). The three light curves shown were some of the earliest DEMONEXT light curves, but are the first for a confirmed exoplanet. The host star has V = 7.58 and the transit depth of ∆V ≈ 14 mmag is clearly detected in the DEMONEXT data. Also visible are the effects of scintillation noise causing photometric accuracy to deteriorate at increasing airmass as the top two transit observations in Figure 3.7 started at high airmass and the point-to-point scatter improves at later observations at lower airmass. These three light curves were taken in May and June 2016 during early commissioning, and are representative of the quality of the earliest DEMONEXT light curves. Since the implementation of science guiding (§ 3.6.2), DEMONEXT routinely achieves 2–4 mmag precision on KELT targets, with mmag precision on data binned on 5 minute timescales.

72 3.7.2. ASAS-SN Monitoring and Conﬁrmation

DEMONEXT has observed 48 ASAS-SN targets to date. Our primary follow-up targets are candidates for supernovae or tidal disruptions events with magnitudes that range from V = 15 at peak and can fade to V ∼ 21 before being removed from the DEMONEXT queue. Typical observations include a sequence of four images in B,V ,R,I with exposure times from 180–300 seconds depending on magnitude. Objects remain in the DEMONEXT queue for around a month, and are observed at a typical cadence of once every 2 nights, although the cadence varies depending on the object.

As an example, Figure 3.8 shows the light curve of SN2016hli, a low-luminosity type Ia supernovae at redshift z = 0.017. SN2016hli was observed by DEMONEXT in B,V,R,I in 20 epochs over 33 nights. The requested cadence was once every 2 nights, but the actual cadence of observations varies, from multiple observations in a single night to ﬁve day gaps, depending on other objects in the queue, demand from KELT targets, and weather. Visible in the DEMONEXT data is the secondary peak for Type Ia in red bands (R,I), that are seen in all but the lowest luminosity Ia. DEMONEXT was still able to detect 2016hli even at V ≈ 20.5.

Attempts were made to confirm targets identified by ASAS-SN in real time using a prompt follow-up tool. In most cases a single-epoch single-band ”raw” image is all that is required to confirm a target for monitoring. Most follow-up observation requests are submitted within a few hours of discovery, and only the initial confirmation image is used. DEMONEXT images are not science ready until noon the next day, so DEMONEXT was unable to submit images before other observatories. A such, we have decided to disable this fast-confirmation observing mode until we can develop a fast reduction system. A serendipitous exception

73 occurred when the author was troubleshooting the telescope during the call for the the conﬁrmation of ASASSN-16fm. We were able to observe and submit a ”raw” image prior to the rest of the community (Villanueva et al. 2016b), demonstrating the utility of rapid conﬁrmation images if we can address the fast reduction issue.

3.7.3. Microlensing Surveys and Follow-Up

In addition to the planned observations of KELT and ASAS-SN targets, DEMONEXT has observed a number of ancillary science programs. Most notable is the observation of microlensing events. DEMONEXT microlensing targets are typically identiﬁed by the OSU users from other surveys such as OGLE (Udalski 2003), MOA (Sako et al. 2008), and KMTNet. The requested observations are of I < 20 objects in low cadence mode at daily cadence to establish baseline observations or to cover the other surveys during times of poor weather. High-cadence mode is used during caustic crossings or other perturbations of interest. In both cases, DEMONEXT has been used to provide data when the other surveys would saturate, typically at V < 13.

DEMONEXT can also be used as a survey telescope. During May and June 2016, DEMONEXT participated in the K2 Campaign 9 microlensing campaign to simultaneously monitor the Galactic Bulge from both the ground and space. DEMONEXT tiled the K2 Campaign 9 super-stamp with a 1 day cadence of 18 ﬁelds in the I band. In the low-cadence mode, DEMONEXT was able to observe all 18 ﬁelds in 2 hours, although the airmass limit was increased from 2.4 to 2.7 to do this. The large 0.5 degree FOV allowed us to tile the entire super-stamp as shown in Figure 3.10. From the reduction of the DEMONEXT Field 18, three microlensing events were observed by DEMONEXT are shown in Figure 3.11. We select these three events to show a brighter event at the top, while progressing to the detection

74 limit of DEMONEXT at the bottom. Although DEMONEXT does not have a suﬃciently high cadence to recover planetary microlensing signatures, it is able to recover microlensing events, even around sources as faint as I ≈ 20, in order to provide robust baseline coverage of these events, which is crucial for modeling any microlensing anomalies.

3.7.4. Open Clusters

Rotation is a fundamental feature of stars, and its evolution throughout stellar lifetimes as a function of mass and age is widely studied (see Bouvier et al. (2014) for a recent review). The primary method for measuring stellar rotation rates is detecting the brightness modulation of stars induced by starspots rotating into and out of the Earth-facing side. The programmatic necessities of these detections, namely high precision and cadence over a baseline of many tens of days, make DEMONEXT a prime instrument for carrying out stellar rotation surveys. As a proof of concept, DEMONEXT has begun systematic monitoring of nearby open clusters with the goal of augmenting the current trove of rotation rates collected in the literature (e.g. Gallet & Bouvier (2015)).

We search for rotation periods using a boot-strap Monte-Carlo analysis of the Lomb-Scargle power spectrum (Lomb 1976; Scargle 1982) of each star following the work of Henderson & Stassun (2012). In this analysis, the observation times are not changed but the magnitude values of the light curve are randomly scrambled. The Lomb-Scargle power is re-calculated during each iteration and if the new power is larger than the original power, the candidate period is rejected. In Figure 3.12, we show an example of a detected rotation rate of 1.689 days with amplitude of 0.013 magnitudes in a possible member of the Alpha Persei cluster. These observations consist of 4,058 individual observations over 12 nights with a 34 day baseline and

75 demonstrate the power of DEMONEXT for uncovering the angular momentum content of nearby stars.

3.8. Chapter Summary

The Dedicated Monitor of Exotransits and Transients (DEMONEXT) is a 0.5 meter robotic telescope equipped with a dedicated 2k2 CCD imager and 10-position filter wheel. Built to observe transiting exoplanets and transients in an automated and efficient way, DEMONEXT has produced 143 light curves for the KELT collaboration, and 48 light curves for the ASAS-SN supernovae group in the first year of operation. In continuous observing mode, DEMONEXT achieves a raw, unbinned, 2–4 mmag precision on bright V < 13 targets with 20–120 second exposures, where most targets are scintillation noise limited. Targets are observed with sub-pixel position stability on the CCD over many hours of continuous observations, and the data can be binned down to 1 mmag precision on 5–6 minute timescales. For single epoch observations, DEMONEXT achieves 1–10% relative photometry on most ASAS-SN targets V < 17 in 5 minute exposures, and is able to detect objects as faint as V ≈ 21. In addition to KELT and ASAS-SN observations, DEMONEXT has also been employed for a number of ancillary science projects, including Galactic microlensing, active galactic nuclei, stellar variability, and stellar rotation, all using the two current observing modes.

76 Fig. 3.1.— DEMONEXT is a PlaneWave CDK20 optical tube assembly on a Mathis Instruments 750/1000 mount pictured at Winer Observatory. DEMONEXT has a 2k2 Finger Lakes Instrumentation CCD, a ten position ﬁlter wheel, and Hedrick electronic focuser. DEMONEXT has a SBIG guide camera and 80 mm guide scope mounted above.

77 X=2.4 X=2.0

X=1.4

X=1.0

Fig. 3.2.— Unbinned precision of DEMONEXT as a function of magnitude for the continuous observation mode. Diagonal dot-dashed line is the standard Poisson signal-to-noise expectation, while the scintillation noise ﬂoor at various airmasses are shown as horizontal dotted lines. Observations were taken at airmasses from X = 1.5–2.4 and the solid line is the X = 2.4 and SNR lines added in quadrature. The gray line is the median in magnitude bins.

78 Fig. 3.3.— Allan variance plot of DEMONEXT data. The precision of the raw, unbinned data is√ the first data point in the top left. The precision of the binned data decreases as t as expected from white noise, but there is marginal evidence of red-noise on long timescales. The line is normalized to the first unbinned data point, but the slope is fixed.

79 Fig. 3.4.— Histogram of measured uncertainties (solid black histogram) compared to the expected uncertainty from photometric and scintillation errors (black dashed curve) with a χ2/dof = 2.89. There are more data points with larger uncertainty than expected due to the systematic errors from ﬂat ﬁelding and fringing which we account for by scaling the errors to achieve a χ2/dof = 1.0 (red histogram).

80 Fig. 3.5.— Histograms of the number of images with the positions of target stars relative to the target stars’ mean for two representative nights. The black (red) curve is a night with good (bad) weather and observing conditions. For the night with good (bad) conditions, there are N = 576 (N = 443) images spanning 5.5 (4.4) hours. The target is kept stable with a median deviation of 0.44 (0.95) pixels.

81 Fig. 3.6.— Histograms of the number of DEMONEXT KELT transit events from 291 nights from May 2016 through June 2017. The gray histogram is the number of transit events scheduled per month, blue were those that were observed, and red are those submitted to the KELT collaboration. Features such as summer shutdown (July– September 2016) are visible. Discrepancies between the black and blue histograms can be attributed to commissioning difficulties (May 2016), weather, or other technical difficulties. Discrepancies between blue and red histograms are attributed to poor weather, poor data quality, or the failure of the filter wheel (March 2017). 76% of observed light curves observed were of sufficient quality to submit to the collaboration, and DEMONEXT averaged 14–15 light curves per month during the first 291 nights of operations.

82 Fig. 3.7.— DEMONEXT observations of KELT-20b/MASCARA-2b (Lund et al. 2017; Talens et al. 2017), the first planet confirmed with DEMONEXT data. DEMONEXT observed three transits from May–June 2016 of the V=7.58 host star with 31 second exposures. The top three light curves (black points) are detrended with respect to the target’s position on the chip and airmass. The light curves have 2–4 mmag RMS, depending on the airmass, relative to the best fit model (red curve). The bottom light curve has all three DEMONEXT light curves phased (gray points). Over-plotted is the phased and binned DEMONEXT light curve (black points) that has 0.89 mmag RMS relative to the model (red curve).

83 Fig. 3.8.— Light curve of type Ia supernoava 2016hli observed by DEMONEXT in 20 epochs. Requested cadence was once every 2 days in B, V , R, and I ﬁlters (blue, green, black, and red points). Visible in the DEMONEXT data is the secondary peak for Type Ia in redder bands (R, I).

84 Fig. 3.9.— Conﬁrmation of ASASSN-16fm taken by DEMONEXT featured in Villanueva et al. (2016b). DEMONEXT images (right) are compared to archival SDSS images (left) to verify the presence of a transient objects.

85 -26 K2C9 Super-stamp DEMONEXT ﬁelds MOA-Ended MOA-Ongoing OGLE-Ended OGLE-Ongoing 'demonex_monitor.txt' u 1:2:3 -27 K2C9F17 K2C9F14

K2C9F15 K2C9F11 K2C9F13 K2C9F08 -28 K2C9F05

K2C9F16 K2C9F10 Decl. (deg) K2C9F12 K2C9F07 K2C9F04 K2C9F02 K2C9F18 K2C9F09 K2C9F06 -29 K2C9F03 K2C9F01

-30 272 271 270 269 268 267 R.A. (deg)

Fig. 3.10.— DEMONEXT tiled the K2 Campaign 9 field in a microlensing campaign to simultaneously monitor the bulge from both the ground and space. DEMONEXT tiled the super-stamp (grey region) with all 18 fields (red areas) in 2 hours each night. Initially reduced fields are outlined in purple, while events detected in other surveys are shown in shades of blue.

86 Fig. 3.11.— Three microlensing events observed by DEMONEXT shown from a robust detection in the top panel, to a marginal detection at the bottom panel. Data from other surveys are in blue and grey, while DEMONEXT data is shown in orange and at lower cadence.

87 Fig. 3.12.— A preliminary light curve of a star with a candidate 1.689 d rotation period. The light curve has been phased on the shown period. The 4058 epochs are shown in light gray, and the data binned into 20 data points is shown in black. The red line denotes the best ﬁt sine curve to the data with amplitude 0.013 magnitudes.

88 Name Model Information

Telescope PlaneWave Instruements CDK20 0.5 m aperature Mount Mathis Instruments MI-750/100 equatorial fork Science Camera Fairchild CCD3041 2k×2k array, 0.9” pix−1, 30.7×30.7 arcmin FOV

89 Filter Wheel Finger Lakes Instruments CFW-3-10 Bessel(Johnson/Cousins) B, V , R, I, Sloan g0, r0, i0, z0, clear Focuser PlaneWave Instruements Hedrick Focuser 33 mm travel distance Guide Telescope Orion 80 mm ED 0.08 m aperature Guide Camera SBIG ST402-ME 765×510 array

Table 3.1. Summary of DEMONEXT parts and components. Software Instruments Controlled

Maxim DL science camera, guide camera, ﬁlter wheel PWI3 focuser Sidereal Technology Servo II mount

Table 3.2. Summary of DEMONEXT software.

90 Chapter 4: Single Transits in TESS

This chapter is based on a submitted draft as seen in Villanueva et al. (2018b).

4.1. Introduction

The Transiting Exoplanet Survey Satellite (TESS), launched in Spring 2018, will discover thousands of transiting exoplanets that exhibit two or more transits during the mission. TESS will have a number of advantages over previous transiting planet surveys, including ground-based surveys such as the Hungarian Automated Telescope (Bakos et al. 2004) survey, the Wide Angle Search for Planets (Pollacco et al. 2006) survey, and the Kilodegree Extremely Little Telescope (Siverd et al. 2012) survey, as well as space-based missions like Corot (Baglin 2003), Kepler (Borucki et al. 2010), and K2 (Howell et al. 2014). Ground-based surveys are essentially limited to planets whose transits have depths above ∼ 0.1%, but do so around bright stars that are amenable to follow-up observations from ground-based telescopes and radial velocity measurements. Kepler has exquisite photometric precision down to several tens of parts-per-million (ppm) (Koch et al. 2010), but the majority of the planets found by Kepler are orbiting stars that are too faint to be conﬁrmed via radial velocity using the current generation of telescopes and instruments. Both the original Kepler campaign and the extended K2 mission campaigns have relatively long baselines of almost 4 years and 80 days respectively, but both are also limited in their sky coverage.

91 By virtue of TESS’s observing strategy and design, it will observe 85% of the entire sky, monitoring and discovering planets transiting bright stars, which are amenable to both photometric and radial velocity follow-up, as well as detailed characterization of their atmospheres via ground and space-based telescopes. The trade-off of achieving this nearly all-sky coverage is that 63% of the sky, or 74% of the mission’s total sky coverage, will only be observed for 27 days (as compared to 80 days for each K2 campaign and nearly 4 years for the primary Kepler campaign). In this regime of many millions of stars monitored for a relatively short amount of time, the number of single-transit planetary events found by TESS will be much larger than that expected or found by Kepler (Yee & Gaudi 2008; Foreman-Mackey et al. 2016). Single transit events require significantly more resources to confirm than planets that exhibit two or more transits, but nevertheless can be quite scientifically valuable.

With the planned survey strategy, and a requirement of at least two transits to confirm a planet, the majority of these planets will have periods of less than 10 days. The primary mission of the survey is to measure masses and radii of 50 terrestrial planets. This leaves open the opportunity to discover planets outside of this regime, including giant planets and planets on long orbits that transit only once. However, recovering, confirming and studying planets that transit only once pose difficulties. Some of these difficulties include the fact that their ephemerides are difficult to constrain for the purpose of scheduling of future observations, and they are easily confused with false-positives.

Previous studies have investigated single-transit events in Kepler (Yee & Gaudi 2008). Multiple simulations have been performed to estimate the (two or more transit) yield of TESS (Sullivan et al. 2015; Bouma et al. 2017; Ballard 2018; Barclay et al. 2018). In each of these simulations, only systems that exhibit two or more

92 transits are reported, and to date there has not been an estimate of the expected yield of single-transits in TESS, or their properties. Given the number of stars and observing strategy of TESS, we expect a 100 fold increase in the number of single-transit events in TESS relative to Kepler.

It is both worthwhile and possible to follow up these longer-period transiting planets, as they represent an opportunity to investigate a number of questions related to planet formation, such as the migration mechanism for Hot Jupiters, and the physical mechanisms that lead to inflated radii of close-in giant planets. Using the definition of habitable zones described by Kopparapu et al. (2013), transiting planets of main sequence stars of spectral type earlier than roughly M5 (Teff ≈ 2800 K) will have the inner edge of the habitable zone at periods of ≈ 11 days. For the majority of the TESS survey, any habitable zone planets around M4 or earlier stars are expected to only display single transits.

4.2. Expected Number of Single-Transit Planets

The expected total number of planets detectable by TESS with exactly one or more transits is the integral over all periods and all planetary radii of the geometric probability of detecting a transit around a star ℘tr, the probability of observing the transit(s) during the ﬁnite baseline of observations ℘B, and the planet occurrence rates f(P ) with a Heaviside step function cut on the signal-to-noise ratio Θ(∆SNR), multiplied by the total number of stars observed by TESS N?

Z Ndet = N? ℘tr℘Bf(P )Θ(∆SNR)dP (4.1)

where ∆SNR = SNR−SNRmin, where each term is for a ﬁxed period and planetary radius. In reality, each of the terms [N?, ℘tr, ℘B, f(P ), Θ(∆SNR)] depend on more

93 than just the period P and planet radius, but also depend on other variables such as the stellar mass, stellar radius, apparent stellar magnitude, and intrinsic stellar variability. All of these variables are considered in the ﬁnal analysis.

We will also assume circular orbits and that rp R?. With this assumption, the geometric transit probability is then

R ℘ = ? (4.2) tr a

for non-grazing geometries, where R? is the host star radius. It is possible to evaluate Equation 4.1 in units of the semi-major axis a, but it is more convenient to use the period P as this is the direct observable in both transit and radial velocity detections of exoplanets. We use Kepler’s third law assuming that the planet’s mass is much smaller than the stellar mass to convert semi-major axis to period

4π2 4π2 P 2 = a3 ≈ a3 (4.3) G(M? + mp) GM? where M? is the stellar mass. The geometric transit probability then becomes a function of stellar mass, stellar radius, and orbital period:

4π2 1/3 ℘ = R M −1/3P −2/3 (4.4) tr G ? ?

−2/3 The geometric probability decreases as ℘tr ∝ P and leads to a decreased probability of detection at long periods.

We also consider the probability of a transit occurring during the ﬁnite baseline of observation B of the TESS mission ℘B. During this paper, we will evaluate cases where two or more transits occur, or exactly one transit occurs. In the case where both two or more transits are observed with planets on periods shorter than B/2, the probability of observing two or more transits occurring during the observing

94 baseline is unity. However for planets on periods longer than B/2, the probability decreases until only one, or no transits occur during the observing baseline. For a ﬁnite observing baseline B and ignoring the ﬁnite duration of the transits, the probability of exactly one ℘B,1, or two or more ℘B,2+ transits occurring is given by:

B ℘B,1 = P ,P ≥ B = 2P − 1 , B ≤ P ≤ B B 2 (4.5) 2P B ℘B,2+ = 2 − B , 2 ≤ P ≤ B B = 1 ,P ≤ 2

These are analogous to Equations 3 and 4 from Yee & Gaudi (2008), but we include the case of two or more transits, instead of exactly two transits. Yee & Gaudi (2008) incorrectly chose the lower limit for the two transit case to be P/4, instead of P/3, however, this only biases their yields for the two-transit cases and their single-transit yields should be unaﬀected. When investigating single-transit events with periods longer than the observing baseline, the total probability is

℘tot ≡ ℘tr℘B R M −1/3 P −5/3 B = 0.026 ? ? (4.6) R M 27.4d 27.4d

−5/3 and scales as ℘tr℘B ∝ P .

The product of these two terms can be seen in Figure 4.1, for representative host stars and a observing baseline of B = 27.4 days. To detect two transits, the probability is the geometric transit probability R?/a until periods of B/2, while the probability of detecting a single transit peaks at B, with the transition for observing

B one versus two transits occurring between 2 ≤ P ≤ B.

95 Each star observed by TESS will have an observing baseline of an integer multiple of N × 27.4 days where 1 ≤ N ≤ 13. The amount of sky covered in each observing baseline is summarized in Table 4.1. The dominant baselines are 73.8% of the mission covered for 27.4 days, 17.8% for 54.8 days, and 3.5% for 82.2 days. There is an uptick at the ecliptic poles, which cover 2% of the mission for 356 days. Each remaining observing baselines cover less than 1% of the mission.

For the total number of stars observed by TESS, we use the TESS Candidate Target List-6 (CTL-6) provided online by Stassun et al. (2017). The catalog has ∼4 million sources, with estimated host star masses, radii, and TESS magnitudes, the host star’s coordinates, and and estimate of the blended flux from nearby stars that is expected to dilute the depth of transits. This method deviates from those used in simulations by Sullivan et al. (2015) and Bouma et al. (2017) as we do not use a Galactic model, but instead calculate our yields directly from the CTL-6. For each of the 4 million stars in the CTL-6, we use the estimated mass, radius, effective temperature, magnitude, and ecliptic latitude of the star. As the the longitude of the first sector was not yet known, we use the ecliptic latitude to assign the number of sectors in which the star will be observed by TESS, and determines the total observing baseline. We do this by taking all stars above a given ecliptic latitude, such that the total area on sky is the same as described in Table 4.1. All observed stars are sorted by the CTL-6 priority, where the top 200,000 stars are classified as postage stamp (PS) stars, and the remaining stars are placed in the FFI sample. Barclay et al. (2018) have show that using only the CTL priority may result in an over-selection of target stars in the ecliptic poles relative to the true mission, however we remain agnostic as to the final selection strategy of the targeted stars and default to CTL-6 priority over speculation as to what the final selected stars will be.

96 We use the planet occurrence rates of Fressin et al. (2013) for stars with

Teff ≥ 4000K, and of Dressing & Charbonneau (2015) for stars with Teff < 4000K. The planet occurrence rates are only complete to periods of ∼ 100 days, but we extrapolate these rates to periods of > 1000 days to explore the probability of finding planets at longer periods. The assumed planet occurrence rates can be seen in Figure 4.3. For each period and radius bin of the respective occurrence rates, we draw a radius and period from a random uniform logarithmic distribution in that bin. From the period of the planet, host star mass, and host star radius we calculate the geometric probability using Equation 4.4. From the ecliptic latitude of the star and the period of the planet, we can calculate the probability of the object being observed for one, two, or more transits from Equation 4.5. We note that, because the probability of detecting a single transit drops precipitously with period (∝ P −5/3), changing our assumed form for the extrapolation in period (within reason) is unlikely to change our results substantially.

We only calculate detections if the SNR is above 7.3. To calculate the SNR, we follow the formula used in Bouma et al. (2017)

p δD SNR = Ntr , (4.7) 2 1/2 σ1hr 2 T + σv

using the number of transits Ntr = 1 for single-transits, the transit depth δ, the dilution from background stars and contamination D bounded from 0–1, the total

2 noise per hour σ1hr from CCD read noise, photon-counting noise, zodiacal noise, and a systematic 60 ppm hr1/2 instrumental noise ﬂoor, the transit duration T , and an intrinsic variability term σv.

We use the host star’s Teﬀ to assign an intrinsic stellar variability based on Basri et al. (2013) following the procedures of Sullivan et al. (2015) and Bouma

97 et al. (2017). Using the planet’s period and radius, along with the host star’s variability and magnitude, we calculate the SNR of each planet, in each period and radius bin, around every star in the sample. For those that have a SNR > 7.3, we deﬁne the planet as being detected. We then sum the product of the geometric probability, probability of being observed, and planet occurrence rate, over all detected planets around all stars. The results are shown in Figure 4.4. The total number of single-transit events is 1218, with 241 of the detections being found in the postage stamps. 201/241 have periods > 25 days, and 19/241 have periods > 250 days. Finally, we recover an estimate of the total integrated number of planets detected. As we integrate fractional probability over all stars, we only recover the total number of planets detected, and not the total number of host stars. As such we and cannot make any quantitative statements on the expected multiplicity of the systems, but one could assume that each star hosts only one planet.

4.2.1. Demographics of Detected Single-Transits

We present the demographics of the detected planets. In Figure 4.5–4.8 we show the distribution of detections in host star magnitude, host star effective temperature, planet radius, and stellar insulation relative to the Earth. Of the 1218 expected single transits, 173 are around stars brighter than T = 10 with 74 around postage stamp stars and 99 in the FFIs. In the postage stamps, the detected planets are split equally 118/124 among cool (Teff < 4000 K) and warm (Teff ≥ 4000 K) stars, but the FFIs favor the warm stars with a 93/883 split between the cool/warm stars. We also

ﬁnd that 196 sub-Neptunes with rp < 4R⊕ will be detected as single-transits in the postage stamps, with an additional 230 detected in the FFIs of CTL-6 stars. All of the planets detected around cool stars have rp < 4R⊕ as there are no planets above

4R⊕ in the Dressing & Charbonneau (2015) occurrence rates. Clanton & Gaudi

98 (2016) show that giant planets do indeed exist around cool stars at long periods, but are uncommon relative to small planets occurrence rates.

Fifty-six planets will be detected in the postage stamps with a stellar insulation within a factor of two of the Earth (0.5 ≤ S/S ≤ 2), and another 73 in the FFIs. Using the conservative habitable zone deﬁned by Kopparapu et al. (2013), we expect 34 habitable zone planets from the postage stamps, 29 are around cool stars, and another 45 habitable zone planets from the FFIs with 32 coming from cool stars.

If we limit the habitable zone planets to terrestrial planets (R ≤ 1.5R⊕) then we expect only 1 planet to be detected, which happens around a postage stamp star with Teff ≤ 4000 K. In a recent simulation of the TESS yield from Barclay et al. (2018), there were no planets found beyond ≈ 85 days which resulted in no planets being found in the habitable zone of FGK stars, where we find 5 in the postage stamps and 13 in the FFIs. It is worth noting that we differ from Barclay et al. (2018) not only in the number of habitable zone planets around stars earlier than M, but also in our target star selection criteria, extrapolation of planet occurrence rates, and in the definition of habitable zone.

We adopt the same SNR threshold SNR = 7.3 as used by both Sullivan et al. (2015) and Bouma et al. (2017) for multiple-transiting events. Given the added uncertainty of single-transit events (e.g. false positives) we follow Barclay et al. (2018) and also look at the distribution of SNR for all of the detections in Figure 4.9 for those wishing to adopt a more stringent SNR cut. We ﬁnd that of the 241 postage stamp detections with SNR ≥ 7.3, 162 have SNR ≥ 10, and 14 have robust detections at SNR ≥ 100. Among the 977 detections around FFI stars with SNR ≥ 7.3, 695 have SNR ≥ 10, and 90 have SNR ≥ 100.

99 4.3. Estimating the Period

An important aspect of identifying the single-transit candidates is the ability to predict the time of future transits to conﬁrm their ephemerides and schedule future observations. One can relate the observed properties of the light curve, the velocity of the planet assuming a circular orbit and zero impact parameter (b = 0), and Kepler’s third law (Eq. 4.3) to relate the period, stellar density, and the light curve observable quantities (Seager & Mall´en-Ornelas2003; Yee & Gaudi 2008; Winn 2010). Beginning with the velocity of the planet

2πa 2R? vp = = , (4.8) P Tdur,0

where Tdur,0 is the duration of the transit at b = 0. One can relate Tdur,0 to the observable quantities Tdur and τ, the measured duration of the transit and ingress/egress time, with

√ T = T 1 − b2 (4.9) dur √dur,0 δT τ = √ dur,0 , (4.10) 1 − b2

√ where δ is the transit depth to arrive at:

2 3/2 Gπ Tdurτ P = ρ∗ √ (4.11) 3 δ

This leads to a degeneracy between the period and the host star density. Seager & Mall´en-Ornelas (2003) showed that it is possible to estimate the period when the mass and radius of the host star is known. Yee & Gaudi (2008) show that the fractional uncertainty in the period (σP /P ) will come from the fractional uncertainty in the density (σρ/ρ) and the fractional uncertainty on the period due to the TESS

100 photometry (σP /P )TESS added in quadrature:

σ 2 σ 2 σ 2 P = ρ + P (4.12) P ρ P TESS

4.3.1. Uncertainty on the Period Due to the Stellar Density

In order to place a constraint on the period, one also needs a constraint on the density of the host star. There are a few avenues that allow for an independent constraint on the density to better than 10%, such that the photometry is the limiting factor of estimating the period of single-transit planets in TESS.

One method to constrain the density will be to ﬁrst estimate the stellar radius

R? of the host star by combining a spectral energy distribution (SED) compiled from broadband photometric measurements or spectrophotometry (when available) with an estimate of the effective temperature Teff of the star. The effective temperature can be obtained from the SED itself or from high-resolution spectra of the host star. By fitting the SED to a model stellar atmosphere, one can estimate the dereddened bolometric flux F? and the extinction (adopting an extinction law). With an estimate of F? and Teff , one can then estimate the stellar angular diameter and thus physical radius of the star using a parallax π from Gaia. The exact precision on R? will depend on the quality of the parallax, SED, and spectra.

With the radius and the surface gravity of the host star log g, the mass and therefore the density of the host star can be estimated. There are a variety of ways one can estimate log g. One can measure this quantity using gravity-sensitive lines in high-resolution spectra, although such spectroscopic estimates of log g can be relatively imprecise, and more importantly, inaccurate, particularly in some regions of parameter space. In some cases, granulation-based ’ﬂicker’ measurements can be

101 used to obtain a more precise estimate of log g (Bastien et al. 2013), however this requires both high quality and relatively long-baseline photometry.

In some cases, it may also be possible to constrain the density from astroseismology. Kjeldsen & Bedding (1995) show that the density scales with the

2 measurable average large-frequency spacing h∆νi as ρ? ∝ h∆νi . In this case, only a 5% measurement of h∆νi is required to place a constraint on the density to 10%. As with ﬂicker measurements, this requires high quality and relatively long baseline photometry.

For most cases, however, we expect that one will ﬁt the radius determined as above, along with metallicity [Fe/H], log g and eﬀective temperature from high-resolution spectra, to stellar isochrones, to determine precise (albeit model- dependent) estimates of the age, mass, and density of the star.

The precision with which the density can be estimated for the star will ultimately depend on which method is used, and the quality of the data being used.

We will simply adopt a ﬁducial value of a 10% precision on ρ?, but note that this may be optimistic in some cases.

In the event that the single-transit planet is part of a multi-planet system with the host star hosting additional interior planets, it will be possible to obtain the density of the host star from transits of the inner planets if any of the inner planets have multiple transits detected and if there is an estimate of their eccentricity. From the period of the inner planets the density can be taken directly from Equation 4.11 and applied as a constraint on the single-transit planet. Ballard (2018) has shown that it is likely that multiple planet systems will be common in TESS, although many will be detected as single-planet systems because the additional planets may go undetected due to lack of SNR or because they only exhibit a single transit.

102 In the end, the exact precision will be determined by which observables are available, and their relative precision. The density constraints of individual stars will likely vary by orders of magnitude, but a 10% precision is expected for many of the brighter, well characterized systems.

4.3.2. Uncertainty on the Period Due to the Photometry

The fractional uncertainty in the period due to the photometry (σP /P )TESS is dominated by the ability to measure the ingress/egress time τ. From Equation 9 in Yee & Gaudi (2008), we get that the fractional uncertainty in the period due to the photometric precision is

σ 2 9 σ 2 1 27T P ≈ τ ≈ dur , (4.13) P TESS 4 τ Q2 2τ and can be related to Q, the approximate total SNR of the transit, and the ratio of the transit duration Tdur to the ingress/egress time τ, assuming τ Tdur. A detailed investigation in to the details of the uncertainties in the observables can be found in Carter et al. (2008) and the Appendix of Yee & Gaudi (2008).

In Figure 4.10 we show the fractional uncertainty expected from single transits based on their photometry. 146/1218 planets will have a fractional uncertainty on the period to better than 10%, with 16 coming from the postage stamps and 130 of those coming from the FFIs. It is worth noting that even in the event of a 1% constraint on the period from photometry, the uncertainty on the density will likely dominate and limit the constraint on the inferred period. Another 72 planets will have a fractional uncertainty on the period of 10-15%, with 5 around postage stamp stars and 72 coming from the FFIs. For cases where the ingress/egress time is shorter than the exposure time, we can only place an upper-limit on the ingress/egress

103 time, and therefore lose the ability to constrain the period. This happens with 373 planets, all identiﬁed in the 30-minute cadence FFIs. The remaining 627 planets all have fractional uncertainties of greater than 15%, where the approximation in Equation 4.13 breaks down and follow-up becomes diﬃcult.

4.3.3. Uncertainty on the Period Due to Eccentricity

Up until now, we have assumed circular orbits for all of the estimates. In reality a number of these planets will likely have non-circular orbits. This will change the duration of the transit and will lead to an incorrect estimation of the period of the planet. Yee & Gaudi (2008) show that the maximum and minimum deviation from the true period is given as:

∆P 1 + e±3/2 = (4.14) P min/max 1 − e

We show this in Figure 4.11 where the minimum and maximum range of true periods of a planet with e = 0.1 would be in the range of a factor of 0.74–1.35 of the assumed circular period. Assuming a median eccentricity of e = 0.17 from the Beta distribution described in Kipping (2013), we get a typical range of periods from 0.59–1.69, or a ≈ 50% uncertainty in the period. This would imply that for non-circular systems, the e 6= 0 uncertainties will limit over our ability to place a constraint using either density or the photometry. However, many of these systems will be in multiple planet systems (Ballard 2018). Zhu et al. (2018) showed that Kepler planet systems become dynamically cooler as the number of planets in the system increases. This follows the results from Xie et al. (2016) that found the mean eccentricity of Kepler multi-planet systems of e = 0.04 to be much lower than that of the single-planet systems e = 0.3. For an eccentricity of e = 0.04, the range

104 of possible periods relative to circular is only 0.89–0.12, which is in line with the expected uncertainty from the density and photometry.

4.4. Prospects for Follow-up

4.4.1. Recovery with Additional Photometry or Precovery in

Archival Data

After estimating the timing of a future transit, we also need to consider which of the single transit candidates will be observable from a typical ground based facility. An additional resource is to look for signals present in existing data sets given a known depth and approximate period. We present the distribution of the undiluted transit depths of single-transits in Figure 4.12. 197/241 planets will be detectable at δ ≥ 0.1% in the postage stamps, with 894/976 of planets detectable around stars in the FFIs. Of these, 40 planets around postage stamp stars will have deep δ ≥ 1% transit depths, and 253 planets around stars in the FFIs.

4.4.2. Expected Radial Velocity Signal

To estimate the expected radial velocity semi-amplitude K, we ﬁrst assign a mass to each planet based on its radius. For planets with radii < 4.0 R⊕ we use the planetary mass-radius relations from Weiss & Marcy (2014) and for planets with radii ≥ 4.0 R⊕ we use the planetary mass-radius relations from Mordasini et al.

(2012). After assigning a planetary mass mp, we use the following equation to assign the radial velocity semi-amplitude K:

−1/3 mp P K = 2/3 (4.15) (mp + M?) 2πG

105 The distribution of expected RV signals from the single transits is shown in Figure 4.13. The majority of single-transit planets 1195/1218, will have RV semi-amplitudes detectable by modern RV instruments K ≥ 1 m/s. 556/1218 will have K ≥ 10 m/s and 20 will have K ≥ 100 m/s.

4.5. Comparison to Other Simulations

As no one has published yields from single transits from TESS simulations, we present the expected yield of the TESS mission proper (i.e. detected two or more transits) for our analysis as a way to compare and scale our results to previous studies. We perform the same analysis described in Section 4.2 to provide an updated estimate of the yield of the primary TESS mission while only considering the number of planets detected with two or more transits at a SNR ≥ 7.3. We ﬁnd 2114 planets detected in the postage stamps and another 5130 in the FFIs around stars in the CTL-6. These can be seen in Figure 4.14. Again, these numbers are incomplete fainter than T > 12 and are lower limits for the FFIs. 255/2114 have periods > 25 days, and only 2/2114 have periods > 250 days in the postage stamps, while and 211/5130 have periods > 25 days and < 1/5130 have periods > 250 days in the FFIs. We also show the yield from single transits as the dashed lines in Figure4.14 for the postage stamps (black) and FFIs (gray). We ﬁnd that within postage stamps the single transits match the TESS mission yield beyond ≈ 25 days, and dominate relative to the expected yield in the FFIs. Beyond ≈ 250 days nearly all detections come from the single-transits in both the postage stamps and FFIs.

We ﬁnd that our predicted yield from the postage stamps of 2114 represents a 20-25% increase over the yield of both the Sullivan et al. (2015) (1734) and Bouma et al. (2017) (1670) estimates using a galactic model, and a 70% increases over the more realistic Barclay et al. (2018) (1250) simulation. The Sullivan et al. (2015)

106 yield predicts >20,000 planets detected in the FFIs, where we only detect 5130 planets in the FFIs, noting that we are incomplete and this is only a lower limit. Bouma et al. (2017) provides a lower limit of 3342 for the FFIs, which is consistent with our estimate. Barclay et al. (2018) found 1250 planets in the postage stamps, with another 3200 planets in the FFIs using the CTL-6 and TIC-6, with another 10,000 planets around stars faint stars not included in the CTL. We ﬁnd that in general we over-estimate the number of planets in the postage stamps relative to other simulations, but are more consistent for the FFIs. It is worth noting that we have extrapolated our planet occurrence rates to much longer periods than all of the above simulations, and have a more simplistic target star selection criteria.

We were able to obtain rough numbers for the estimated number of single transit events from individual trials of other simulations via private communication. Again, we ﬁnd 977 single-transits in the FFIs and 241 in the postage stamps in our work. From one trial from the Sullivan et al. (2015) simulation, we estimate ≈ 1300 single-transits in the FFIs and ≈ 100 single-transits in the postage stamps (P. Sullivan, private communication). From one trial from the Bouma et al. (2017) simulation, we estimate ≈ 850 single-transits in the FFIs and ≈ 150 single-transits in the postage stamps (L. Bouman, private communication). Although unpublished, we ﬁnd that the total number of single-transit events from each simulation, including our work, varies from 1000–1400 with each group disagreeing on the relative fraction found in the postage stamps versus the FFIs.

4.6. Recommendations for Observations

Given that nearly all of the 1218 single-transit events detectable in TESS will have either photometric (90%) or radial velocity (98%) signals measurable from current ground-based observatories, there will be more planets detected than could possibly

107 be followed-up. Follow-up observers should coordinate to prioritize which planets will be targeted for follow-up observations, either by their ability to constrain the period, signal-to-noise ratio, or scientiﬁc merit. With 98% of planets detectable in RV, that the RV measurements are required to determine the mass and planetary nature of the planets, and that RV measurements can help constrain the period, RV resources should be immediately allocated towards conﬁrming single-transit events. RV will be crucial to the single transit detections in single-planet systems with poorly constrained eccentricities. Additionally, for those with photometric signals, searches in archival data and planned observations around the predicted next transit can be used to determine the period and constrain the timing of future transits.

4.7. Chapter Summary

The number of single-transit planets from TESS is expected to be an order of magnitude greater than those found in Kepler, with 241 single-transit planets detected in the postage stamps, and another 977 detected from the FFIs around stars brighter than T = 12. Single transits require greater follow-up resources than the typical TESS planet, and there will be more single-transit planets signals than follow-up resources will be able to observe or conﬁrm. This is despite the fact that 90% and 98% of all such planets detected will have photometric and RV signals respectively that will be observable from current ground-based observatories.

It is possible to predict future transits of single-transits by placing constraints on the light curve observables and on the density of the stellar host. The uncertainties from the density of the host star will be ≈ 10% in many cases, however only 10% of the planets (146) will have photometry suﬃcient to provide constraints on the period to better than a 10% due to uncertainties on the photometry assuming circular orbits. The uncertainty due to eccentric orbits will make constraining the

108 true period diﬃcult, but multi-planet systems represent the best systems to place constraints on both the stellar density and eccentricity.

Our single-transit yields predict a 80% increase in the number of planets detected beyond 25 days compared to the TESS mission, and a factor of 12 increase in the yield for planets beyond 250 days. This includes 79 habitable zone planets and ∼ 1 terrestrial planet in the habitable zone. This opportunity to substantially augment the yield of the TESS mission should not be overlooked. However, given the“abundance of riches” represented by these single-transit events, we recommend community collaboration to make the most of these opportunities.

109 Fig. 4.1.— Probability of observing a single transit (solid) or two or more transits (dashed) for the 27.4 day baseline as compared to the geometric transit probability (dotted). Colors correspond to a 1.0, 0.6, and 0.25 M host star.

110 Fig. 4.2.— Mission-weighted probability of observing a single transit (solid black line) or two or more transits (dashed black line) over all observing baselines as compared to the geometric transit probability (red dotted line) for a 1.0 M host star. All 13 individual single transit probability curves, corresponding to the 13 possible baselines, are shown as grey solid lines for reference.

111 Fig. 4.3.— Planet occurrence rates for Teﬀ ≥ 4000 K stars (blue) and Teﬀ < 4000 K stars (red). Dark lines are the fraction of stars expected to host at least one planet in each period bin, while the dotted lines represent our extrapolation to long periods.

112 Fig. 4.4.— Combining Figures 4.1–4.3 with the total number of stars, we estimate the period distribution of single transit events expected from TESS postage stamps (black) and in the FFIs (gray). Events around stars with Teﬀ ≥ 4000K are in blue, and stars Teﬀ < 4000 K in red. The darker shades are for the 2-minute cadence, while the lighter shades are for the 30-minute cadence. The total number of planets exhibiting a single transit event expected from the TESS mission is over 1000. There are 241 planets expected in the 2-minute cadence data, and lower limit of 977 planets in the FFIs.

113 Fig. 4.5.— Number of expected single transit events by magnitude. Dotted gray line is the demographics of the CTL-6 normalized to fit on this scale. The solid black and gray histograms are the total yield from the postage stamps and FFIs respectively. These are subdivided into Teff ≥ 4000 K (blue) and Teff < 4000 K (red) star samples with the darker shades for postage stamps and the lighter shades for the FFIs.

114 Fig. 4.6.— Number of expected single transit events by Teff . Dotted gray line is the demographics of the CTL-6 normalized to fit on this scale. The solid black and gray histograms are the total yield from the postage stamps and FFIs respectively. These are subdivided into Teff ≥ 4000 K (blue) and Teff < 4000 K (red) star samples with the darker shades for postage stamps and the lighter shades for the FFIs.

115 Fig. 4.7.— Number of expected single transit events by planet radius. Dotted gray line is the demographics of the CTL-6 normalized to fit on this scale. The solid black and gray histograms are the total yield from the postage stamps and FFIs respectively. These are subdivided into Teff ≥ 4000 K (blue) and Teff < 4000 K (red) star samples with the darker shades for postage stamps and the lighter shades for the FFIs.

116 Fig. 4.8.— Number of expected single transit events by stellar insulation relative to the Sun. Dotted gray line is the demographics of the CTL-6 normalized to fit on this scale. The solid black and gray histograms are the total yield from the postage stamps and FFIs respectively. These are subdivided into Teff ≥ 4000 K (blue) and Teff < 4000 K (red) star samples with the darker shades for postage stamps and the lighter shades for the FFIs.

117 Fig. 4.9.— The number of single transit detections by SNR with a SNR threshold of SNR = 7.3. Colors are the same as in Figure 4.5. 162/241 postage stamp star detections have SNR ≥ 10, and 14/241 have SNR ≥ 100. 695/977 detections around FFI stars have SNR ≥ 10, and 90/977 have SNR ≥ 100.

118 Fig. 4.10.— Number of expected single transit events by photometric uncertainty on the period. Colors are the same as in Figure 4.5. The separate bin to the far right are the objects where the ingress/egress time was shorter than the exposure time and therefore only a upper limit can be place on the ingress/egress time.

119 Fig. 4.11.— The maximum and minimum deviation from the true period under the assumption of a circular orbit. The spread is due to the change in planet velocity and transit duration when the planet transits at periastron versus apastron.

120 Fig. 4.12.— Expected single transit transit depths. Colors are the same as in Figure 4.5. Of planets detected, 90% and 24% of the planets will have transit depths deeper than 0.1% and 1% respectively.

121 Fig. 4.13.— Expected single transit transit radial velocity signals. Colors are the same as in Figure 4.5. Of planets detected, 98% and 46% of the planets will have RV signals greater than 1 and 10 m/s respectively.

122 Fig. 4.14.— Expected yield from the TESS mission. Colors are the same as in Figure 4.5. Single transit distribution from Figure 4.4 are shown in dashed lines. Single transits in the postage stamps match the TESS mission yield for planets with 25 ≤ P ≤ 250 days, and dominate relative to the expected yield in the FFIs. Nearly all planets detected with P ≥ 250 come from the single-transits in both the postage stamps and FFIs.

123 Days Square Degrees Sky Fraction Mission Fraction

0 6023 14.6 0 27.4 25989 63.0 73.8 54.8 6270 15.2 17.8 82.2 1238 3.0 3.5 110 231 0.56 0.66 137...301 578(total) 1.4(total) 1.6(total) 329 215 0.52 0.6 356 701 1.7 2.0

Table 4.1. Fraction of sky covered by various observing baselines.

124 Chapter 5: Conclusions and Future Work

5.1. Summary

We have shown through these three publications, that it is possible to precisely measure the masses, radii, and a host of planetary and stellar properties through observations of transiting Hot Jupiters combined with additional data. Understanding the relationship between a star and planet’s masses and its radii, is the ﬁrst step in understanding the system and its properties. We have also shown that it is possible to observe these Hot Jupiters in a automated and routine way that provides a large quantity of high quality data. Finally, we provide a potential source of thousands of Hot Juptier progenitors to study, identiﬁed as single-transits in TESS.

5.2. Future Work

Building oﬀ of this work with future collaborators, we will identify and recover candidate single-transit planets in TESS, vet the candidates for false-positives, estimate their periods, and schedule follow-up observations. Single epoch imaging from seeing-limited telescopes on the ground, like DEMONEXT, will conﬁrm the presence of nearby neighbours and blends in the TESS images to eliminate eclipsing binaries and eliminate those that have depths too deep to be caused by planets. Single transit events require a constraint on the stellar density to estimate the period, and we will make use of information about the stellar host, using reconnaissance

125 spectroscopy, and archival data, to place constraints on the stellar density to allow period estimation to within a 10% uncertainty. This results in the ability to predict future transits to within a few days or weeks.

We will confirm the timing of transits with the 0.5-meter robotic DEMONEXT, that is capable of producing large numbers of precision transit light curves of giant planets around bright targets. The timing of smaller planets, and the masses of both large and small planets, will be confirmed with radial velocity resources. By coordinating additional follow-up resources, it will be possible to confirm the planetary nature of some of brightest, longest period, and interesting exoplanets to date.

By providing a sample of bright, long-period, and potentially habitable planets, we give astronomers the opportunity to understand exoplanets in a new regime. These planets would have been missed by TESS, are in an interesting regime where the planets are unaﬀected by stellar irradiation, but recoverable with DEMONEXT. Understanding the masses, radii, and demographics of these non-inﬂated planets, places stronger constraints on planet occurrence and formation models than will be achieved by the primary TESS mission, and gives DEMONEXT a plethora of targets to observe over the next three years.

126 References

Albrecht, S., Winn, J. N., Johnson, J. A., et al. 2012, ApJ, 757, 18A

Alonso, R., Brown, T. M., Torres, G., et al. 2004, ApJL, 613, L153

Auvergne, M., Bodin, P., & Boisnard, L. 2009, A&A, 506, 411

Baglin, A. 2003, AdSpR, 31, 345B

Bakos, G., Noyes, R. W., Kov´acs,G., et al. 2004, PASP, 116, 266B

Barclay, T., Pepper, J., & Quintana, E. V. 2018, arXiv:1804.05050

Ballard, S., 2018, arXiv:1801.04949

Basri, G., Walkowicz, L. M., Reiners, A., 2013, ApJ, 769, 37B

Bastien, F. A., Stassun, K. G., Basri, G., & Pepper, J. 2013, Nature, 500, 427

Benedict, G. F., McArthur, B. E., Gatewood, G., et al. 2006, AJ, 132, 2206

Benomar, O., Masuda, K., Shibahashi, H., & Suto, Y. 2014, PASP, 66, 94B

Bond, I. A., Udalski, A., Jaroszy`nsky, M., et al. 2004, ApJ, 606, L155

Borucki, W. J., Koch, D., Basri, G., et al. 2010, Sci, 327, 977B

Bou´e,G., Montalto, M., Boisse, I., Oshagh, M., & Santos, N. C. 2013, A&A Rev., 550A, 53B

Bouma, L. G., Winn, J. N., Kosiarek, J., & McCullough, P. R., 2017, arXiv:1705.08891

Bourrier, V., Lecavelier des Etangs, A., H´ebrard,G., et al. 2015, 579A, 55B

Bouvier, J., Matt, S. P., Mohanty, S., et al. 2014, Proc. Protostars and Planets VI, 433B

Brown, T. M., Baliber, N., Bianco, F. B., et al. 2013, PASP, 125, 1031B

127 de Bruijne, J. H. J. 2012, Ap&SS, 341, 31D

Campbell, B., Walker, G. A. H., & Yang, S. 1988, ApJ, 331, 902

Carter, J. A., Yee, J. C., Eastman, J., et al. 2008, ApJ, 689, 499C

Charbonneau, D., Brown, T. M., Latham, D. W., et al. 2000, ApJ, 529, L45

Clanton, C. & Gaudi, B. S. 2016, ApJ, 819, 125C

Claret, A. & Bloemen, S. 2011, A&A Rev., 529A, 75C

Collier Cameron, A., Bruce, V. A., Miller, G. R. M., et al. 2010, MNRAS, 403, 151C

Collins, K. A., Eastman, J. D., Beatty, T. G., et al. 2014, AJ, 147, 39C

Demarque, P., Woo, J., Kim, Y., & Yi, S. K. 2004, ApJS, 155, 667D

Demory, B.-O. & Seager, S., 2011, ApJS, 197, 12D

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

Eastman, J., Gaudi, B. S., Siverd, R., et al. 2010, Proc. SPIE, 7733E, 3JE

Eastman, J. D., Siverd, R., & Gaudi, B. S. 2010, PASP, 122, 935E

Eastman, J. D., Gaudi, B. S., & Agol, E. 2013, PASP, 125,83-112

Eastman, J. D., Beatty, T. G., Siverd, R., et al. 2015, arXiv:1510.00015

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

Fazio, G. G., Hora, J. L., Allen, L. E., et al. 2004, ApJS, 154, 10F

Foreman-Mackey, D., Morton, T. D., Hogg, D. W., et al. 2016, AJ, 152, 206F

Fressin, F., Torres, G., Charbonneau, D., 2013, ApJ, 766, 81F

Gallet, F. & Bouvier, J. 2015, A&A, 577A, 95G

Gaudi, B. S., Chang, H. Y., & Han, C. 2003, ApJ, 586, 527g

Gaudi, B. S. & Winn, J. N. 2007, ApJ, 655, 550G

Gillon, M., Pont, F., Moutou, C., et al. 2007, A&A, 466, 743G

Gillon, M., Jehin, E., Lederer, S. M., et al. 2016, Nature, 533, 221G

Gim´enez,A. 2006, ApJ, 650, 408G

128 Hartman, J. D., Stanek, K. Z., Gaudi, B. S., Holman, M. J., & McLeod, B. A. 2005, AJ, 130, 2241

Henderson, C. B. & Stassun, K. G. 2012, ApJ, 747, 51H

Henry, G. W., Marcy, G. Butler, R. P., et al. 1999, IAU Circ., 7307, 1

Hirano, T., Suto, Y., Taruya, A., et al. 2010, ApJ, 709, 458

Hirano, T., Suto, Y., Winn, J., et al. 2011, ApJ, 742, 69H

Howell, S. B., Sobeck, C., Haas, M., et al. 2014, PASP, 126, 398H

Irwin, J., Charbonneau, D., Nutzman, P., & Falco, E. 2009, IAU Symp., 253, 37

Ivezic, Z., Tyson, J. A., Abel, B., et al. 2008, http://arxiv.org/abs/0805.2366 (Accessed: 22 June 2017)

Johnson, J. A., Winn, J. N., Narita, N., et al. 2008, ApJ, 686, 649J

Kipping, D. 2013, MNRAS, 434L, 51K

Kjeldsen H. & Bedding T. R. 1995, A&A, 293, 87K

Koch, D. G., Borucki, W. J., Basri, G., et al. 2010, ApJ, 713, L79-L86

Konacki, M., Torres. G., Jha, S. et al. 2003, Nature, 421, 507

Kopparapu, R. K., Ramirez, R., Kasting, J. F., et al. 2013, ApJ, 765, 131K

Kub´anek,P. 2008, www.rts2.org

Kub´anek,P., Falco, E., Jel´ınek,M., et al. 2012, Proc. SPIE, 8448E, 11K

Kuhn, R. B., Rodriguez, J. E., Collins, K. A., et al. 2016, MNRAS, 459 4281K

Lagrange, A., Kasper, M., Boccaletti, A., et al. 2009, A&A, 506, 927

Latham, D. W., Stefanik, R. P., Mazeh, T. et al. 1989, Nature, 339, 38L

Lomb, N. R. 1976, Ap&SS, 39, 447L

Lund, M. B., Rodriguez, J. E., Zhou, G., et al. 2017, 2017arXiv170701518L

Marcy, G., Butler, R. P., Fischer, D., et al. 2005, PThPS, 158, 24M

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

McCullough, P. R., Stys, J. E., Valenti, J. A., et al. 2005, PASP, 117, 783

129 McCullough, P. R., Stys, J. E., Valenti, J. A., et al. 2006, ApJ, 648, 1228

McCullough, P. R., Burke, C., Valenti, J. A., et al. 2008, arXiv:0805.2921

McLaughlin, D. B. 1924, ApJ, 60, 22M

Mordasini, C., Alibert, Y., Georgy, C., et al. 2012, A&A, 547A, 112M

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

Narita, N., Hirano, T., Sato, B., et al. 2009, PASJ, 61, 991

Narita, N., Hirano, T., Sanchis-Ojeda, R., et al. 2010, PASJ, 62, L61-L65

Nutzman, P. & Charbonneau, D. 2008, PASP, 120, 317N

Ohta, Y., Taruya, A., & Suto, Y. 2005, ApJ, 622, 1118

Ohta, Y., Taruya, A., & Suto, Y. 2009, ApJ, 690, 1

Pepper, J., Pogge, R. W., DePoy, D. L., et al. 2007, PASP, 119, 923P

Pepper, J., Kuhn, R. B., Siverd, R., et al. 2012, PASP, 124, 230P

Pollacco, D. L., Skillen, I., Collier Cameron, A., et al. 2006, PASP, 118, 1407P

Pont, F., Zucker, S., & Queloz, D. 2006, MNRAS, 373, 231P

Ricker, G. R., Winn, J. N., Vanderspek, R., et al. 2015, JATIS, 1, a4003R

Rossiter, R. A. 1924, ApJ, 60, 15R

Sako, T., Sekiguchi, T., Sasaki, M., et al. 2008, ExA, 22, 51S

Scargle, J. D. 1982, ApJ, 263, 835S

Seager, S. & Mall´en-Ornelas,G., 2003, ApJ, 585, 1038S

Shappee, B. J., Prieto, J. L., Grupe, D., et al. 2014, ApJ, 788, 48S

Siverd, R. J., Beatty, T. G., Pepper, J., et al. 2012, ApJ, 61, 123S

Stassun, K. G., Oelkers, R. J., Pepper, J., et al. 2017, arxiv:1706.00495

Sullivan, P. W., Winn, J. N., Berta-Thompson, Z. K., et al. 2015, ApJ, 809, 77S

Swift, J. J., Bottom, M., Johnson, J. A., et al. 2015, JATIS, 1b7002S

Talens, G. J. J., Justesen, A. B., Albrecht, S., et al. 2017, 2017arXiv170701500T

130 Temple, L. Y., Hellier, C., Albrow, M. D., et al. 2017, 2017arXiv170407771T

Thorngren, D. P., Fortney, J. J., Murray-Clay, R. A., & Lopez, E. D. 2016, ApJ, 831, 64T

Todorov K. O., Deming, D., Knutson, H. A., et al. 2012, ApJ, 746, 111

Torres, G., Andersen, J., & Gim´enez,A. 2010, A&ARv, 18, 67T

Queloz, D., Eggenberger, A., Mayor, M., et al. 2000, A&A, 359, 13

Udalski, A. 1992, Acta Astron., 53, 291U

Villanueva, S., Jr., Eastman, J. D., & Gaudi, B. S. 2016, ApJ, 820, 87V

Villanueva, S., Jr., Eastman, J. D., Gaudi, B. S., et al. 2016 Proc. SPIE, 9906E, 2LV

Villanueva, S., Jr., Brown, J. S., Stanek, K. Z., et al. 2016, ATel, 9080, 1V

Villanueva, S., Jr., Gaudi, B. S., Pogge, R. W., et al. 2018, PASP, 130, 5001V

Villanueva, S., Jr., Dragomir, D., & Gaudi, B. S. 2018, 2018arXiv180500956V

Werner, M. W., Roellig, T. L., Low, F. J., et al. 2004, ApJS, 154, 1W

Weiss, L. M. & Marcy, G. W. 2014, ApJ, 783L, 6W

Winn, J. N., Noyes, R. W., Holman, M. J., et al. 2005, ApJ, 631, 1215

Winn, J. N., Johnson, J. A., Narita, N., et al. 2008, ApJ, 682, 1283W

Winn, J., 2010, arXiv:1001.2010

Winn, J. N., Fabrycky, D., Albrecht, S., & Johnson, J. A. 2010, ApJ, 718L, 145W

Wolszczan, A. & Frail, D. A. 1992, Nature, 335, 145

Xie, J., Dong, S., Zhu, Z., et al. 2016, PNAS, 11311431X

Yee, J. C. & Gaudi, B. S. 2008, ApJ, 688, 616Y

Yi, S., Demarque, P., Kim, Y., et al. 2001, ApJS, 136, 417Y

Young, A. T. 1967, AJ, 72, 747

Zhu, W., Petrovich, C., Wu, Y., et al. 2018, arXiv:1802.09526

131