On the Atmospheres of the Smallest Gas Exoplanets
by
Erin M. May
A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Astronomoy and Astrophysics) in The University of Michigan 2019
Doctoral Committee: Assistant Professor Emily Rasucher, Chair Professor Fred Adams Professor Michael Meyer Professor John Monnier
Erin M. May [email protected] ORCID iD: 0000-0002-2739-1465 c Erin M. May 2019 For my cat. May she learn to love me at least as much as I love this dissertation.
ii ACKNOWLEDGEMENTS
“I don’t have emotions. And sometimes that makes me very sad.” – Bender, the Robot
I’ve never been much for emotions, but if it’s a required component of this dissertation...
First, thank you to Alex. I may have been a pain to deal with during parts of this process, but we both made it through. Thank you to Li’l B who taught me that not everything that’s perfect is perfect for you and that love isn’t unconditional. Especially from a cat.
Thank you to the humans in the department who were there for me along the way. In particular, Emily, who was the advisor I needed but didn’t deserve.
To Renee for confirming that there’s no such thing as too many macarons on a Friday, and for the constant commiserating throughout the past year. And because this human requested this acknowledgement, to Adi for saving me that one time from that one thing.
Thank you to the regular GB@3 crew, even those who showed up late or barely at all. Thank you to Danilyn who made GB feel like a home. Also to Brian who always knew that the only thing that mattered was tacos and how many we wanted (needed).
Thank you to all the other humans, mostly those who were especially nice humans. You know who you are.
Finally, to the humans who told me I couldn’t do it, thank you for giving me someone to prove wrong. Because, well, I did it.
iii TABLE OF CONTENTS
DEDICATION ...... ii
ACKNOWLEDGEMENTS ...... iii
LIST OF FIGURES ...... viii
LIST OF TABLES ...... xi
ABSTRACT ...... xiii
CHAPTER
I. Introduction ...... 1
1.1 Detection of Exoplanets ...... 1 1.1.1 The Transit Method ...... 2 1.1.2 The Radial Velocity Method ...... 12 1.1.3 Current Status of Exoplanet Detection and the Future of the Field ...... 16 1.2 Basic Parameters and Classifications of Exoplanets ...... 17 1.3 Transmission Spectroscopy ...... 21 1.3.1 Space-based Observations ...... 25 1.3.2 Ground-based Observations ...... 26 1.4 Atmospheric Modeling ...... 27 1.4.1 Energy Balance Models ...... 28 1.4.2 General Circulation Models ...... 29 1.5 Dissertation Overview ...... 30
II. MOPSS I: Flat Optical Spectra for the Hot Jupiters WASP-4b and WASP-52b ...... 31
2.1 Preface ...... 31 2.2 Abstract ...... 31 2.3 Introduction ...... 32
iv 2.4 Observations ...... 34 2.4.1 The IMACS Instrument ...... 34 2.4.2 The planet WASP-4b ...... 36 2.4.3 The planet WASP-52b ...... 36 2.5 Data Analysis ...... 38 2.5.1 Reduction Pipeline ...... 38 2.5.2 Light Curves ...... 42 2.5.2.1 White Light Curves ...... 43 2.5.2.2 Binned Light Curves ...... 44 2.5.3 Error Analysis and Noise Removal ...... 50 2.6 Results ...... 51 2.6.1 Transmission Spectra ...... 51 2.6.2 WASP-4b ...... 52 2.6.3 WASP-52b ...... 54 2.6.3.1 Unocculted Star Spots ...... 54 2.6.3.2 Star Spots on the Stellar Limb ...... 55 2.7 Conclusions ...... 59 2.8 Acknowledgements ...... 60 2.9 Facilities and Software Used ...... 60 2.10 Appendix ...... 61 2.10.1 Check Star ...... 61 2.10.2 Tables of Results ...... 63
III. MOPSS II: Extreme Optical Scattering Slope for the Inflated Super- Neptune HATS-8b ...... 67
3.1 Preface ...... 67 3.2 Abstract ...... 67 3.3 Introduction ...... 68 3.4 Observations ...... 71 3.4.1 The Inamori-Magellan Areal Camera & Spectrograph In- strument ...... 71 3.4.2 The planet HATS-8b ...... 72 3.5 Data Analysis ...... 73 3.5.1 Reduction Pipeline ...... 73 3.5.2 Removing Airmass Trend ...... 74 3.5.3 Correlated Noise Model ...... 75 3.5.4 Light Curves ...... 76 3.5.4.1 White Light Curves ...... 77 3.5.4.2 Binned Light Curves ...... 78 3.6 Results ...... 79 3.6.1 Transmission Spectrum ...... 79 3.6.2 Unocculted Star Spots ...... 84 3.6.3 A Clear Atmosphere ...... 88 3.6.4 Clouds and the Scattering Slope ...... 88
v 3.7 Conclusions ...... 92 3.8 Acknowledgements ...... 93 3.9 Facilities and Software Used ...... 93
IV. Living On the Edge: The Effects of a Surface on Atmospheric Cir- culation and Emission for 1.5R⊕ Planets ...... 95
4.1 Preface ...... 95 4.2 Abstract ...... 95 4.3 Introduction ...... 96 4.4 Method ...... 98 4.4.1 General Circulation Model ...... 98 4.4.2 Surface-Atmosphere Interactions ...... 99 4.4.2.1 Surface Heating ...... 100 4.4.2.2 Atmospheric Drag ...... 101 4.4.3 Choice of Model Resolution ...... 102 4.5 Model Validation ...... 103 4.5.1 Earth ...... 103 4.5.1.1 Model Inputs ...... 104 4.5.1.2 Results: Earth ...... 106 4.6 Transition Planet ...... 106 4.6.1 Planetary Properties ...... 108 4.6.2 Model Iterations ...... 110 4.7 Results: Transition Planet ...... 111 4.7.1 Band Integrated Emission and Top of the Atmosphere Albedos112 4.7.2 Zonal Averaged Emission and Reflection ...... 113 4.8 Conclusions and Future Directions ...... 117
V. Examining Tatooine: Atmospheric Models of Neptune-like Cir- cumbinary Planets ...... 118
5.1 Preface ...... 118 5.2 Abstract ...... 118 5.3 Introduction ...... 119 5.4 Method ...... 121 5.4.1 Calculation of Orbits and Resulting Irradiation ...... 121 5.4.2 Energy Balance Model ...... 122 5.4.2.1 C: Heat Capacity Per Unit Area ...... 124 5.4.2.2 D: Diffusivity ...... 124 5.4.2.3 S(1-A): Stellar Irradiation and Albedo . . . . . 125 5.4.2.4 I: Cooling Function ...... 126 5.4.3 General Circulation Model ...... 126 5.4.3.1 GCM Heat Transport - applications to the EBM 128 5.5 Results ...... 128 5.5.1 Kepler 47b ...... 129
vi 5.5.1.1 Kepler 47b - EBM ...... 129 5.5.1.2 Kepler 47b - GCM ...... 131 5.5.2 Expanding Models to More Planets ...... 134 5.5.3 Introduction of Obliquity ...... 136 5.6 Conclusions ...... 139 5.6.1 Modeling of Circumbinary Planets ...... 139 5.6.2 Habitability of Circumbinary Planets ...... 140 5.6.3 Observables ...... 140
VI. Conclusion ...... 142
6.1 Summary ...... 142 6.1.1 Transmission Spectroscopy and the Future of Exoplanet Ob- servations ...... 142 6.1.2 Atmospheric Modeling and the Future of Atmospheric Char- acterization ...... 144 6.2 Future Directions ...... 145
BIBLIOGRAPHY ...... 148
vii LIST OF FIGURES
Figure
1.1 Schematic of a Transit ...... 3
1.2 Geometry of a Transit ...... 4
1.3 Schematic of Limb Darkening ...... 7
1.4 Limb Darkening and its effect on transits ...... 8
1.5 Occulted Spots and their effect on transits ...... 10
1.6 Exoplanets discovered with the transit method to-date...... 11
1.7 Radial Velocity data of 51 Pegasi b ...... 13
1.8 All exoplanets discovered to-date...... 16
1.9 Mass vs. Radius for known Exoplanets ...... 19
1.10 Composition Curves with Known Exoplanets Plotted ...... 20
1.11 Transmission Spectroscopy Schematic ...... 22
2.1 Field of Views for WASP-4b and WASP-52b Observations...... 35
2.2 WASP-4 Spectrum ...... 37
2.3 WASP-52 Spectrum ...... 39
2.4 White light curve and baseline models for WASP-4b ...... 43
2.5 White light curve for WASP-52b ...... 45
viii 2.6 Binned light curves for WASP-4b ...... 46
2.7 Binned light curves for WASP-52b ...... 48
2.8 Limb Darkening results for WASP-4b and WASP-52b ...... 50
2.9 Transmission Spectra for WASP-4b ...... 53
2.10 Combined Transmission Spectra for WASP-52b ...... 54
2.11 The effects of star spots on measured transmission spectra ...... 56
2.12 The effects of occulted spots on the stellar limb on the transit light curve . 57
2.13 The effects of occulted spots on the stellar limb on the fit limb darkening parameters ...... 57
2.14 The effects of occulted spots on the stellar limb on the fit limb darkening parameters for WASP-52b specifically ...... 58
2.15 Check-star Spectrum ...... 61
2.16 Check-star reduction process ...... 62
3.1 Field of View for HATS-8 ...... 72
3.2 Airmass correction for HATS-8b observations ...... 75
3.3 Correlated Noise Model for HATS-8b ...... 77
3.4 Light curves for HATS-8b ...... 81
3.5 Limb Darkening Fits from HATS-8b ...... 82
3.6 Transmission Spectrum for HATS-8b ...... 83
3.7 Effects of Unocculted Spots on the Transmission Spectrum ...... 87
3.8 Necessary atmospheric parameters to explain the measured scattering slope for HATS-8b in a clear atmosphere ...... 89
3.9 Necessary atmospheric parameters to explain the measured scattering slope for HATS-8b in a cloudy atmosphere ...... 91
4.1 Earth GCM - Resolutions dependencies of temperature ...... 107
ix 4.2 Earth Temperature-Pressure Profile Comparison ...... 108
4.3 Definitions of Transition Regime Base GCM Models ...... 109
4.4 Long wave to Short Wave Comparison ...... 112
4.5 Emitted Flux for 1.5R⊕ Planets ...... 114
4.6 Equator-to-Pole Emitted Long Wave Flux Differences ...... 116
5.1 Incident flux variations for representative circumbinary systems...... 123
5.2 Time variable temperature patterns as predicted by the EBM for Kepler-47b.131
5.3 Temperature and wind maps as predicted by the GCM for Kepler-47b with both time-varying and constant irradiation inputs...... 132
5.4 Expected temperature fluctuations for differing levels of modeling complex- ities for Kepler-47b...... 133
5.5 Variability of known circumbinary planetary atmospheres as compared to their single star cases...... 136
5.6 Variability of a grid of circumbinary systems as compared to their single star cases, for a variety of axial obliquites...... 137
5.7 Time variable temperature patterns as predicted by the EBM for Kepler-47b with a 30◦ obliquity...... 138
x LIST OF TABLES
Table
1.1 Current Status of MOPSS ...... 27
2.1 WASP-4b: Calibrator Stars ...... 37
2.2 WASP-52b: Calibrator Stars ...... 39
2.3 WASP-4b Stellar and Orbital Parameters ...... 49
2.4 WASP-52b Stellar and Orbital Parameters ...... 49
2.5 WASP-4b: Wavelength dependent variables ...... 63
2.6 WASP-52b: Wavelength dependent variables, ut20160811 ...... 64
2.7 WASP-52b: Wavelength dependent variables, ut20160922 ...... 65
2.8 WASP-52b: Wavelength dependent variables, Combined Rp/Rstar values . . 66
3.1 HATS-8b Calibrator Stars ...... 73
3.2 HATS-8b Stellar and Orbital Parameters ...... 78
3.3 HATS-8b: MCMC Fit Results ...... 86
4.1 Surface Heat Equation Parameters ...... 103
4.2 Earth GCM Input Parameters ...... 105
4.3 GCM Parameters for set of 4 Base Models ...... 111
5.1 Circumbinary Exoplanet Energy Balance Model Parameters ...... 126
xi 5.2 Circumbinary Exoplanets General Circulation Model Parameters ...... 127
5.3 Kepler 47 System Parameters ...... 130
xii ABSTRACT
With nearly 4000 unique exoplanets detected to-date, the field of exoplanet characteriza- tion has numerous questions to answer about these diverse planets and plenty of targets to observe as we work to understand the population of exoplanets as a whole, and where Earth
1 fits in. In particular, planets larger than Earth but smaller than Neptune (3.88 R⊕ ) are the most common exoplanet detected, yet we lack a direct comparison in our Solar System to study and extrapolate from. Because there must be some point at which exoplanets transition from having a higher likelihood of being terrestrial (thin gaseous envelope compared to the total radius) to having a higher likelihood of being gaseous (large gaseous envelope compared to the total radius), this size-regime has been dubbed the ‘transition regime’. With this in mind, in this dissertation, I outline the most prominent current detection methods and how the information we gain from them, specifically mass and radius, informs our understanding of detected exoplanets both within and outside the transition regime. I further discuss how these parameters alone are not enough to classify planets as terrestrial or gaseous within the transition regime due to degeneracies in inferred composition from mass-radius relations. In this dissertation I take a two-pronged approach to work towards a better understanding of the classification of these transition planets through both direct observations and theory. First, in Chapters II and III I present my work to directly measure the compositions of the smallest observable gas planets using ground-based transmission spectroscopy. This tech- nique, while not immediately probing the transition regime, is a step towards small-planet characterization as instrumentation and observational techniques improve. In Chapter IV I use a three-dimensional general circulation model to study the influence of solid surfaces on atmospheric circulation for transition regime planets and study how the surface imprints
1 Where R⊕ refers to the radius of Earth
xiii itself on the emitted and reflected light from the exoplanet with an eye toward observing these differences. In Chapter V, I further explore the atmospheres of small, Neptune-sized planets in circumbinary orbits with time varying incident flux patterns due to their orbits around their pair of host stars to explore how this environment impacts the circulation. Together, these approaches are unique, yet complementary, in their contribution towards understanding the population of exoplanets as a whole.
xiv CHAPTER I
Introduction
“There are countless suns and countless earths all rotating round their suns in exactly the same way as the seven planets of our system.” – Giordano Bruno
1.1 Detection of Exoplanets
Four centuries after the philosopher Giordano Bruno famously suggested that every star in the night sky might be host to planets of their own, the first objects of confirmed planetary masses were detected around the pulsar PSR B1257+12. Dubbed PSR B1257+12 B and PSR
B1257+12 C, these planets were small at 4.3 M⊕ and 3.9 M⊕, respectively, and skirted the very line of terrestrial planets vs. gaseous planets. (Wolszczan & Frail, 1992). Though they orbited a ‘dead’ star, these first two exoplanets were representative of what we would come to learn are the most common types of exoplanets around main sequence stars (Batalha, 2014). Now, almost 30 years after the detection of the pulsar planets, just over1 4000 exoplanets have been officially confirmed (NASA Exoplanet Archive2). The detection of these exoplanets
1I’m so excited that we hit #Exoplanet4k between writing and turning in this dissertation! 2exoplanetarchive.ipac.caltech.edu
1 is dominated by two techniques: (1) transits of the exoplanet across the disk of the host star to enable measurements of the exoplanet’s radius; and (2) radial velocity measurements of the host star to enable measurements of the exoplanet’s mass. Here I outline the science behind these detection techniques as well as their limitations and successes to-date before discussing the current and future status of the subfield of exoplanet detection.
1.1.1 The Transit Method
Events such as the Transit of Venus in 2012 and even the North American Total Solar Eclipse in 2017 grab the public’s attention and demonstrate the broad reach of astronomical transit observations. As applied to exoplanet science, transit observations have been a key component dictating the fast expansion of our knowledge of other worlds. Simply put, a transit event occurs when an orbiting body, such as an exoplanet, crosses in front of the host star as viewed along an observer’s line-of-sight. During such an event, a portion of the stellar light is blocked from view resulting in the star appearing to be dimmer in the sky for the duration of the transit event. If this occurs on a regular, predictable, schedule, one can surmise that a body is in a orbit around the studied star. Figure 1.1 demonstrates a schematic of a transit event. As shown, the exoplanet blocks out an amount of light proportional to the ratio of the area of the exoplanet disk to stellar disk, and to first order the relative dimming of the star is given by
2 Rp f = 2 (1.1) R?
where f is the fractional decrease in detected light; Rp is the radius of the exoplanet; and
R? is the radius of the host star. If we know the stellar radius, we can derive the exoplanet radius3. For the same size host star, larger exoplanets create larger transits and smaller exoplanets create smaller transits.
3Of course, the measurement of exoplanet radius is only as precise as our knowledge of the stellar radius. And so, here one could go on a long aside titled, as many a conference and tweet has been, “To know thy planet, one must first know thy star”. But alas, I leave this as an exercise to the reader.
2 Figure 1.1: A schematic representation of a transit.
While analytical forms of the transit light curve equation can be derived (see P´al,2008), here we walk through the derivation of the integral form, typically solved numerically. For a radial symmetric intensity profile (with the radial dimension in units of R?=1), the total flux during a point of time during transit can be written as
1 2π Z Z Ftransit(t) = I(r)dφrdr 0 2δ(r,t) (1.2) 1 Z = 2 (π − δp(r, t)) I(r)rdr 0
where δ is half of the arc obscured by the exoplanet (see Figure 1.2, panel a), I(r) is the intensity profile of the star (for a uniform disk, this would be constant with r, see below for a discussion of I(r), namely limb darkening and its effects on transits). Note that the angular
integral runs from 2δp(r) through 2π, which covers the entire radial arc while skipping the
region obscured by the exoplanet. Figure 1.2 panel b shows the angle δp(r) specifically for
r = R?, but it can be generalized for any arc of radius r such that
2 2 2 r + zp − Rp δp(r, t) = arccos (1.3) 2rzp(t)
3 Figure 1.2: A schematic of various transit geometries highlighting relevant angles.
where r is the radial direction of integration; Rp is the exoplanet radius; and zp is the distance from the center of the star to the center of the exoplanet, all defined in a coordinate system centered on the star where R?=1.
p 2 2 To calculate zp(t) where zp(t) = xp(t) + yp(t) and xp(t) is the planet’s motion in the plane of its orbit and yp(t) is the planet’s motion perpendicular to the plane of its orbit, we first follow the derivation for planar coordinates in P´al(2009), adapted from Murray & Dermott (1999) which define
ξ = ξ0 cosω ¯ − η0 sinω ¯ (1.4)
η = ξ0 sinω ¯ + η0 cosω ¯ (1.5) with
ξ0 = a(cos(E − e)) (1.6) √ 2 η0 = a 1 − e sin E (1.7) where ξ is the projected vertical motion (motion perpendicular to the plane of the orbit); ω¯and η is the projected horizontal motion (motion in the plane of the orbit);ω ¯ is the longitude of pericenter; E is the eccentric anomaly; and e is the eccentricity of the orbit.
4 From Kepler’s equation, E is given by
E − e sin E = M (1.8)
where M is the mean anomaly, defined by 2πt/T , with t the time and T the orbital period. The mean longitude of the orbit is defined as λ = M +ω ¯. For circular (or nearly circular) orbits,ω ¯ = 0, one can write Equations 1.6 and 1.7 as
ξ0 = a cos λ (1.9)
η0 = a sin λ (1.10)
noting, that in this case, ξ and η reduce to ξ0 and η0, respectively. Finally, one can non- dimensionalize the vertical and horizontal elements of the planet’s motion in units of stellar radii as bξ y = (1.11) p a η xp = (1.12) R?
where a is the semi-major axis of the orbit and b is the impact parameter (defined in Figure
1.2 panel c) given by b = a cos i/R?, with i the inclination of the orbit. Equations 1.9-1.10 and 1.11-1.12 then reduce to
2 2 2 2 2πt a 2 2πt zp = b cos + sin (1.13) T R? T
which now allows us to calculate δ(r, t) (Equation 1.3), and integrate over the radial axis to
obtain Ftransit(t). As one can surmise, a variety of system parameters (e.g. semi-major axis, orbital period, orbital inclination, exoplanet radius, stellar radius) play an important role in the determi- nation of the transit light curve. Now I’ll look specifically at the stellar intensity profile
5 component, I(r). Because the host star is in reality a non-uniform disk, its intensity profile adds complexity to transit studies, both with the treatment of limb darkening and due to the presence of star spots on the disk4. Limb darkening is a geometric effect resulting in the observer receiving photons from different depths in the stellar photosphere at different annuli of the observed stellar disk. As described by panel a in Figure 1.3, the τ = 1 surface probes different depths in the photosphere at the outer radial regions of the star (i.e. larger φ angles) than the central regions of the star (lower φ angles). Because photons emitted at the same radial distance from the stellar center, but at varying values of φ, have different amounts of material to pass through on their way to the observer, those at larger values of φ will be absorbed before those at smaller values of φ. At larger values of φ we then detect photons from the higher/cooler regions in the photosphere from which photons are able to escape before being absorbed and which appear darker than the deeper/warmer regions. This is a wavelength dependent effect wherein limb darkening appears stronger at shorter (bluer) wavelengths. This effect impacts transits by resulting in a lower relative amount of stellar flux being blocked by the exoplanet as it passes either limb during ingress and egress as compared to the central regions of the stellar disk. As demonstrated by Figure 1.4, with all else the same, the shape of the bottom of the transit is determined by the strength of limb darkening; stronger limb darkening (i.e. the limb is much darker than the central regions of the star) corresponds to a smoother transition from out-of-transit through ingress/egress. In contrast, weaker limb darkening creates a sharper transition. The radial intensity profile of the stellar disk due to limb darkening can be parameterized in a number of ways with varying levels of complexity and accuracy. Commonly in transit
4Once again, “To know thy planet, one must first know thy star”
6 Figure 1.3: A schematic representation of limb darkening. a: showing the τ=1 surface effect. b: showing the geometry of the parameterized limb darkening laws.
observations limb darkening is parameterized as one of three relations given by
I(µ) = I0 [1 − c1 (1 − µ)] (linear law) (1.14)
2 I(µ) = I0 1 − c1 (1 − µ) − c2 1 − µ (quadratic law) (1.15)
1/2 I(µ) = I0 1 − c1 1 − µ − c2 (1 − µ)