Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Corvus (constellation) wikipedia , lookup
Aquarius (constellation) wikipedia , lookup
International Ultraviolet Explorer wikipedia , lookup
Astrophotography wikipedia , lookup
Extraterrestrial life wikipedia , lookup
Interferometry wikipedia , lookup
Timeline of astronomy wikipedia , lookup
The Pennsylvania State University The Graduate School Department of Electrical Engineering CHARACTERIZING PROPERTIES OF OPTICAL FIBERS AND FIBER COUPLINGS TO NEAR-INFRARED HIGH-RESOLUTION SPECTROGRAPHS CRITICAL TO PRECISION RADIAL-VELOCITY STUDIES A Thesis in Electrical Engineering by Keegan S. McCoy 2012 Keegan S. McCoy Submitted in Partial Fulfillment of the Requirements for the Degree of Master of Science May 2012 The thesis of Keegan S. McCoy was reviewed and approved* by the following: Lawrence Ramsey Professor of Astronomy and Astrophysics Thesis Advisor Timothy Kane Professor of Electrical Engineering Sven Bilén Associate Professor of Engineering Design, Electrical Engineering, and Aerospace Engineering Kultegin Aydin Professor of Electrical Engineering Head of the Department of Electrical Engineering *Signatures are on file in the Graduate School iii ABSTRACT The search for exoplanets, planets orbiting stars beyond our own Solar System, is one of the fastest-growing and most exciting areas of astronomical research. Pathfinder, the Penn State Astronomy and Astrophysics Department's fiber-fed, near-infrared (NIR) spectrograph, is designed to explore the technical issues that must be resolved in order to measure precise radial velocities of M dwarfs from ground-based observations at the Hobby-Eberly Telescope. These stellar radial velocities are induced in host stars, such as M dwarfs, by the gravitational interaction with exoplanets. Pathfinder already has demonstrated <10 m/s short-term radial-velocity precision in the NIR, but this thesis investigates the visible and NIR properties of the optical fibers that couple light from the telescope to Pathfinder to achieve even higher precision. Since the stability of a spectrograph's instrument profile is crucial to accurately measure stellar radial velocities, a series of tests were performed on fused-silica, multimode fibers to measure and mitigate modal noise, the time variation of an optical fiber's modal power distribution. The lessons learned from this research and the Pathfinder prototype will be used in the Habitable Zone Planet Finder (HPF), which is a future facility-class, high-resolution NIR spectrograph that will be capable of high-precision (<5 m/s) radial-velocity measurements of Earth-mass planets orbiting mid- to late-type M dwarfs. iv TABLE OF CONTENTS LIST OF FIGURES ................................................................................................... vi LIST OF TABLES ..................................................................................................... xi ACKNOWLEDGMENTS ......................................................................................... xii Chapter 1 Introduction ............................................................................................ 1 1.1 History of Exoplanet Detection ..................................................................... 2 1.1.1 Early Exoplanet Claims ...................................................................... 2 1.1.2 Confirmed Discoveries ....................................................................... 3 1.2 The Radial-Velocity Method of Detecting Exoplanets ................................. 4 1.2.1 Physical Detection Mechanism .......................................................... 5 1.2.2 Spectral Information ........................................................................... 10 1.2.3 Stellar Noise ....................................................................................... 13 1.3 Exoplanet Habitability ................................................................................... 14 1.3.1 The Habitable Zone ............................................................................ 16 1.3.2 Exoplanet Atmospheres ...................................................................... 19 1.4 M Dwarfs as Exoplanet Host Stars ................................................................ 22 1.4.1 Properties of M Dwarfs ...................................................................... 22 1.4.2 M Dwarf Exoplanets .......................................................................... 27 1.4.2.1 M Dwarf Activity ................................................................ 28 1.4.2.2 Tidal Locking and Synchronous Rotation ........................... 31 1.5 Current Challenges of Near-Infrared Precision Radial-Velocity Measurements ................................................................................................ 32 1.5.1 NIR Calibration Sources .................................................................... 33 1.5.1.1 I2 Gas Cell ........................................................................... 33 1.5.1.2 Th–Ar Lamp ........................................................................ 34 1.5.1.3 Laser Frequency Comb ....................................................... 35 1.5.2 Telluric Absorption ............................................................................ 36 1.5.3 Optical Fibers ..................................................................................... 37 1.5.3.1 Modal Description ............................................................... 38 1.5.3.2 Modal Noise and Scrambling .............................................. 40 1.6 Current NIR Instruments ............................................................................... 45 1.7 Conclusions ................................................................................................... 46 v Chapter 2 Penn State Pathfinder............................................................................. 47 2.1 2.2 2.3 2.4 Spectrograph ................................................................................................. 48 NIR Detector ................................................................................................. 51 Calibration System ........................................................................................ 52 Preliminary Results from the Hobby-Eberly Telescope ............................... 54 Chapter 3 Preliminary Laser Diode Modal Noise Tests........................................ 59 3.1 Test Procedure .............................................................................................. 59 3.2 Results and Analysis ..................................................................................... 62 3.3 Optical System and Image Analysis Improvements ..................................... 67 3.3.1 Camera Calibration ........................................................................... 68 3.3.2 Dust Mitigation ................................................................................. 70 3.3.3 Diffraction Effects ............................................................................ 71 3.3.4 Optical Coupling ............................................................................... 72 3.4 Conclusions ................................................................................................... 74 Chapter 4 Visible and Near-Infrared Modal Noise Tests ..................................... 76 4.1 Laser Diode and Narrow-Band Laser Tests .................................................. 76 4.1.1 Test Procedure .................................................................................. 76 4.1.2 Results and Analysis ......................................................................... 79 4.1.3 Conclusions ....................................................................................... 103 Chapter 5 Conclusions and Future Work .............................................................. 106 References ................................................................................................................... 109 vi LIST OF FIGURES Figure 1-1: An exoplanet and its host star orbiting their common center of mass, denoted by the red plus sign (figure not to scale). ................................................ 6 Figure 1-2: Sample radial-velocity curve for a star with an orbiting exoplanet, with period P. The exoplanet induces a radial-velocity semi-amplitude, K, in the host star's motion (Clubb 2008). ..................................................................... 7 Figure 1-3: Definition of orbital inclination. When i = 90° the orbit is seen edgeon, and when it is 0° it is seen face-on (Casoli and Encrenaz 2007). ................... 7 Figure 1-4: Exoplanet minimum mass v. semi-major axis for planets detected using the RV method. The mass of Earth is highlighted by the red line and the green circle is located at Earth's orbital distance from the Sun. ..................... 9 Figure 1-5: Radial velocity Doppler shift in a host star's spectrum. ........................... 10 Figure 1-6: Solar spectrum with absorption lines (Pogge). ........................................ 11 Figure 1-7: The zero-age main sequence habitable zone. Stellar classification is provided on the vertical axis and the dotted line indicates the tidal locking radius after 4.5 Gyr (Kasting et al. 2003). ............................................................ 17 Figure 1-8: Schematic of the habitable zone (green zone) of different host stars (not to scale). "Hotter" stars roughly translates to O, B, and A stars, "Sunlike" stars to F and G stars, and "Cooler" stars to K and M stars (not to scale) ("Kepler Artwork"). .................................................................................... 17 Figure 1-9: Population of known stars within 10 parsecs (~32.6 light years), binned by spectral type. ........................................................................................ 23 Figure 1-10: Evolution of several low-mass M dwarfs on the Hertzsprung-Russell diagram. The inset diagram displays the main-sequence lifetimes as a function of mass (Laughlin et al. 1997). ............................................................... 25 Figure 1-11: Comparison of stellar spectral evolution (Engle and Guinan 2011). ..... 26 Figure 1-12: Spectral energy distribution for an F, G, and M star (Kasting et al. 1993). .................................................................................................................... 27 Figure 1-13: Activity lifetime of M dwarfs (West et al. 2008). .................................. 29 vii Figure 1-14: Fraction of active M dwarfs as a function of subclass (West et al. 2008). .................................................................................................................... 29 Figure 1-15: Evolution of low-mass M dwarf habitable zone boundaries (Khodachenko et al. 2007). ................................................................................... 30 Figure 1-16: Atmospheric transmission in the NIR (Geballe 2010). .......................... 37 Figure 1-17: Mode propagation through an optical fiber (Corbett and AllingtonSmith 2006). ......................................................................................................... 39 Figure 1-18: Effect of fiber non-uniformity on modal propagation (Corbett and Allington-Smith 2006). ......................................................................................... 40 Figure 1-19: Number of modes in the far-visible and NIR for a multimode optical fiber (200-µm-diameter core and 0.22 NA). ......................................................... 41 Figure 1-20: The position shift of a point source at the fiber input corresponds to a linear PSF shift on the detector (Avila and Singh 2008). .................................. 44 Figure 1-21: Operation of a double scrambler. The top panel (a) shows that light exiting the output end of the input fiber at a particular position enters the input face of the output fiber at a particular angle. The lower panel (b) shows that light exiting the output end of the input fiber at a particular angle enters the input end of the output fiber at a particular position (Hunter and Ramsey 1992). .................................................................................................................... 45 Figure 2-1: Schematic layout of the Pathfinder spectrograph (Ramsey et al. 2010). .................................................................................................................... 49 Figure 2-2: Simultaneous exposure of science and calibration fibers (Ramsey et al. 2010). ............................................................................................................... 50 Figure 2-3: Pathfinder on the optical bench during an engineering test run at HET. Light from the HET and the calibration system is fiber-fed into a slit mechanism (1), bounces off a gold fold mirror (2), reflects off the collimating mirror (3), is dispersed off the grating (covered, 4), is cross-dispersed off the cross disperser (5), reflects off the camera mirror (6), and reflects off a gold mirror (not visible, 7) into the dewar, where approximately 0.2% of the light from the telescope lands on the detector (8). This layout is very similar to the optical diagram shown in Figure 2.1, except in the latter case the fiber input mechanism is at the location of the above-pictured gold mirror (2) (Redman 2011).. ................................................................................................................... 51 Figure 2-4: Free spectral range of the orders in the Y- and J-bands for Pathfinder's cross-dispersed echelle configuration, along with the coverage viii for our current detector (same size as an H1R) and an H2RG, which will be used in HPF (Ramsey et al. 2010).. ...................................................................... 52 Figure 2-5: Comparison of U and Th line densities in the Y-band (Ramsey et al. 2010). .................................................................................................................... 54 Figure 2-6: Velocity variation for HD106714 over the May 2010 test run (Ramsey et al. 2010). ............................................................................................ 55 Figure 2-7: Top: The three 300-μm fibers, separated by 125-μm (outer diameter) fibers that acted as spacers for the larger fibers. This fiber pseudo-slit was assembled and polished on-site. From left to right, these are the secondary calibration, primary calibration, and science (stellar) fibers. Bottom: The same fibers, re-imaged onto a 100-μm slit. In this image, the primary calibration fiber is transmitting light. The slit was aligned to optimize the orientation of the primary calibration and HET fibers (Redman 2011). .............. 57 Figure 2-8: Schematic of coupling calibration sources into Pathfinder (Ycas et al. 2012). .................................................................................................................... 58 Figure 2-9: Spectra of the star HD16823 along with LFC calibration. Echelle orders 37–39 are visible, with wavelengths of 1536.6–1543.9 nm, 1577.1– 1584.6 nm, and 1619.7–1627.4 nm, respectively (Ycas et al. 2012).................... 58 Figure 3-1: Xenics NIR camera quantum efficiency. ................................................. 60 Figure 3-2: Vibrator attached to 200-µm multimode fiber. ........................................ 61 Figure 3-3: 635-nm single mode fiber test: (left) 635-nm single mode fiber (average of 10 frames), (right) 3D intensity plot. ................................................. 62 Figure 3-4: 635-nm static and vibrated fiber tests: (left) 635-nm static (average of 10 frames), (right) 635-nm vibrated (average of 10 frames). ........................... 63 Figure 3-5: 1550-nm static and vibrated fiber tests: (left) 1550-nm static test (average of 10 frames), (right) 1550-nm vibrated test (average of 10 frames). ... 64 Figure 3-6: 2D spatial frequency power spectra: 635-nm static (solid blue line), 635-nm vibrated (solid red line); 1550-nm static (dashed blue line), 1550 nm vibrated (dashed red line). .................................................................................... 67 Figure 3-7: Dolan-Jenner area backlight used for flat fielding. .................................. 69 Figure 3-8: Improved flat field frame. ........................................................................ 69 Figure 3-9: Normalized master flat field frame. ......................................................... 70 ix Figure 3-10: Sample image highlighting dust accumulation. ..................................... 71 Figure 3-11: Diffraction effects caused by motorized iris. ......................................... 72 Figure 3-12: Single mode fiber optically coupled to the multimode fiber. ................ 73 Figure 3-13: Schematic of optical coupling setup within the overall test configuration. ........................................................................................................ 74 Figure 4-1: Test setup for comprehensive laser and laser diode tests......................... 77 Figure 4-2: FLI PL4710 deep depletion camera quantum efficiency (black line). ..... 78 Figure 4-3: 635-nm static fiber tests: (top left) Xenics camera image, (top right) Xenics camera image 2D FFT, (bottom left) FLI camera image (bottom right) FLI camera image 2D FFT. .................................................................................. 80 Figure 4-4: 635-nm laser diode vibration tests (FLI camera): (top left) mechanical vibration image, (top right) mechanical vibration image 2D FFT, (bottom left) hand agitation image, (bottom right) hand agitation image 2D FFT. ...................................................................................................................... 83 Figure 4-5: 635-nm laser diode tests power spectra (FLI camera): (blue line) static image power spectrum, (black line) mechanical vibration image power spectrum, (red line) hand agitation image power spectrum. ................................. 84 Figure 4-6: 632.8-nm HeNe laser tests (FLI camera): (top left) static image, (top right) static image 2D FFT, (middle left) mechanical vibration image, (middle left) mechanical vibration image 2D FFT, (bottom left) hand agitation image, (bottom right) hand agitation image 2D FFT. ............................ 86 Figure 4-7: 632.8-nm HeNe laser tests power spectra (FLI camera): (blue line) static image power spectrum, (black line) mechanical vibration image power spectrum, (red line) hand agitation image power spectrum. ................................. 87 Figure 4-8: 830-nm laser diode tests (FLI camera): (top left) static image, (top right) static image 2D FFT, (middle left) mechanical vibration image, (middle left) mechanical vibration image 2D FFT, (bottom left) hand agitation image, (bottom right) hand agitation image 2D FFT. ............................ 89 Figure 4-9: 830-nm laser diode tests power spectra (FLI camera): (blue line) static image power spectrum, (black line) mechanical vibration image power spectrum, (red line) hand agitation image power spectrum. ................................. 90 Figure 4-10: 1310-nm laser diode tests (Xenics camera): (top left) static image, (top right) static image 2D FFT, (middle left) mechanical vibration image, x (middle left) mechanical vibration image 2D FFT, (bottom left) hand agitation image, (bottom right) hand agitation image 2D FFT. ............................ 92 Figure 4-11: 1310-nm laser diode tests power spectra (Xenics camera): (blue line) static image power spectrum, (black line) mechanical vibration image power spectrum, (red line) hand agitation image power spectrum. ...................... 93 Figure 4-12: 1550-nm laser diode tests (Xenics camera): (top left) static image, (top right) static image 2D FFT, (middle left) mechanical vibration image, (middle left) mechanical vibration image 2D FFT, (bottom left) hand agitation image, (bottom right) hand agitation image 2D FFT. ............................ 94 Figure 4-13: 1550-nm laser diode tests power spectra (Xenics camera): (blue line) static image power spectrum, (black line) mechanical vibration image power spectrum, (red line) hand agitation image power spectrum. ...................... 95 Figure 4-14: 1550-nm narrow-band laser tests (Xenics camera): (top left) static image, (top right) static image 2D FFT, (middle left) mechanical vibration image, (middle left) mechanical vibration image 2D FFT, (bottom left) hand agitation image, (bottom right) hand agitation image 2D FFT. ............................ 97 Figure 4-15: 1550-nm narrow-band laser tests power spectra (Xenics camera): (blue line) static image power spectrum, (black line) mechanical vibration image power spectrum, (red line) hand agitation image power spectrum. ........... 98 Figure 4-16: Laser and laser diode test speckle contrasts: (blue) static images, (green) mechanical vibration images (red) hand agitation images. “LD” indicates laser diode, "F" and "X" denote FLI and Xenics cameras, respectively, and "Laser633" specifies the HeNe laser. ....................................... 102 Figure 4-17: Preliminary mechanical design of the large-amplitude agitator. ........... 105 xi LIST OF TABLES Table 1-1: Radial velocity comparison table for various planets orbiting a solarmass star (Pasquini 2009). .................................................................................... 9 Table 1-2: Physical characteristics of each M dwarf subclass (Kaltenegger and Traub 2009). ......................................................................................................... 24 Table 1-3: Orbital distance of equivalent solar flux for M dwarfs (solar flux = 1360 W/m2) (Heath et al. 1998). ........................................................................... 26 Table 3-1: Thor Labs laser diodes used in preliminary modal noise testing. ............. 60 Table 4-1: Laser and laser diode specification (LD = laser diode). ............................ 77 Table 4-2: Speckle contrasts for laser and laser diode tests........................................ 101 xii ACKNOWLEDGMENTS First and foremost, I would like to thank God for being an undying source of love, strength, courage, and wisdom, not just during my six years at Penn State, but throughout my entire life. Studying science and engineering has truly shown me the amazing work of your power and beauty. I continue to grow in my faith each and every day, and I am grateful for our eternal walk together. Much gratitude is extended to Dr. Larry Ramsey, my thesis advisor, for teaching me how to be both an engineer and a scientist and for providing me with much encouragement during our time together the past few years. It has been a pleasure to work with someone who is so knowledgeable and experienced in the world of astronomical instrumentation. I am appreciative of Dr. Suvrath Mahadevan for his guidance in helping to design our test configurations and providing the necessary equipment. Sam Halverson deserves special thanks for his time and immense patience assisting me in performing the tests presented in this thesis. Much of this work would not have been possible without his support. I wish to thank all of the children who I have had the opportunity to teach during my planetarium and Mars 3D shows, classroom visits, and other science outreach events over the years. You remind me of the pure joy and mystery of science through your enthusiasm and eagerness to learn. In many ways, you have taught me more than I have taught you. xiii To my parents and my entire family, I cherish all of the love and support that you have provided me during my college years. I have been truly blessed and I thank you each one of you for believing in me every step of the way. To my grandpa, Ken Zimmerman, I thank you for instilling in me a love of astronomy which I will hold for the rest of my life. To the U.S. Air Force, I not only thank you for the financial support that you have provided me the past two years, but also for the tremendous honor of serving the United States as an Air Force officer. From an early age, I have fostered a deep respect for our military and all of its great heroes and leaders throughout our history. I look forward to my career as a developmental engineering officer and I solemnly swear to serve with honor, distinction, and a love of all that our country holds dear. To Major Richard "Dick" Winters, former commander of Company E, 2nd Battalion, 506th Parachute Infantry Regiment, 101st Airborne Division, you will forever be my hero and I only hope that I live up to be half the man that you were, not just as an officer, but in life. Finally, I humbly dedicate this thesis to all of the brave men and women of the U.S. Armed Forces who have given the last full measure of devotion in the service of our country. From this day to the ending of the world, But we in it shall be remember'd; We few, we happy few, we band of brothers. – King Henry V, from William Shakespeare's Henry V xiv The heavens declare the glory of God; The skies proclaim the work of his hands. Psalms 19:1 (NIV) 1 Chapter 1 Introduction The question of whether or not there exist worlds beyond our own is an age-old inquiry intricately interwoven with philosophy, theology, imagination, and science. In 4th century B.C. ancient Greece, the philosopher Epicurus believed that there were an infinite number of worlds, while Aristotle hypothesized that the Earth was unique and at the center of the Universe (Casoli and Encrenaz 2007). It was Aristotle's view of an Earth-centric Universe that would dominate scientific thinking for the next 1,800 years. In 1543, the Polish mathematician and astronomer Nicolas Copernicus radically challenged this paradigm, when he postulated that the Sun was at the center of the Universe with the Earth and all the other planets in orbit around it. This fundamental change in thinking, with the Earth no longer holding its special universal location, led the more imaginative thinkers of the day to consider if other planets orbited the multitude of stars in the night sky. Giordano Bruno, an Italian Dominican friar and philosopher, was one such brave soul. In his book, De L'infinito Universo E Mondi, published in 1584, Bruno wrote (Kasting 2010): There are countless suns and countless earths all rotating around their suns in exactly the same way as the seven planets of our system . . . The countless worlds in the universe are no worse and no less inhabited than our Earth. 2 Though Bruno would ultimately be burned at the stake for a number of his heretical statements, such as this one, the idea of worlds beyond our own gained more traction as Copernicus' heliocentric Solar System model garnered widespread support over the following centuries. The goal of this chapter is to provide an overview of the search for exoplanets, planets orbiting stars beyond our Solar System, from both an engineering and planetary science perspective. Understanding the technical intricacies and astronomical context of M dwarf exoplanet detection is crucial before discussing Penn State Pathfinder and its role in detecting potentially habitable, Earth-mass exoplanets orbiting M dwarfs. 1.1 History of Exoplanet Detection Christiaan Huygens, the prominent Dutch mathematician and astronomer, was the first person to attempt direct observation of exoplanets in the late 17th century, but he quickly realized that the task was impossible with the instruments available to him at the time (Casoli and Encrenaz 2007). However, as the capability of astronomical instrumentation increased over the years, the possibility of detecting exoplanets slowly began to move from a dream to reality. 1.1.1 Early Exoplanet Claims By the 19th century, astronomers were beginning to find evidence of the presence of smaller companions to stars based on the minute periodic motions of those stars in 3 relation to other nearby stars. The first claim of an exoplanet detection came in 1855, when Captain William S. Jacob observed the binary star 70 Ophiuchi at the East India Company's Madras Observatory. Captain Jacob concluded that reported orbital anomalies in the 70 Ophiuchi system made it "highly probable" that there was, indeed, a "planetary body in connection with this system" (Jacob 1855). However, over 40 years later, American astronomer Forest Ray Moulton published a paper based on the solution of the restricted three-body problem, proving that a third body would make the 70 Ophiuchi system highly unstable (Moulton 1899). In 1963, after several other unconfirmed reports of stellar companions, Dutch astronomer Peter van de Kamp of Swarthmore College's Sproul Observatory drew considerable attention for his alleged discovery of a planet with 1.6 times the mass of Jupiter orbiting Barnard's Star (van de Kamp 1963). Unfortunately, this result was also later refuted, when studies by Gatewood and Eichhorn (1973) and Hershey (1973) revealed that systematic errors in the Sproul Observatory telescope were responsible for the astrometric shifts. Once again, the discovery of an exoplanet proved elusive. 1.1.2 Confirmed Discoveries The breakthrough finally came in 1992, when Wolszczan and Frail (1992) reported the discovery of two planets orbiting the pulsar PSR1257+12. Considered the first definitive detection of exoplanets, the discovery was made by observing slight variations in the periodic arrival of pulsar radio pulses throughout the orbital periods of the two planets. In 1995, Mayor and Queloz (1995) announced the detection of the first 4 exoplanet orbiting a main-sequence star, namely the G-type star 51 Pegasi. With half of a Jupiter mass and orbiting only 0.05 astronomical units (AU) from its host star, 51 Pegasi b was the first of many "hot Jupiters" to dominate the early exoplanet discoveries. Requiring only a few days worth of observation due to its short orbital period, astronomers had never suspected that a planet could orbit so close to its host star. As of March 2012, 763 exoplanets have been detected orbiting other stars1 using a variety of techniques. NASA's Kepler mission, which is currently observing over 160,000 stars in the constellation of Cygnus, has also identified an additional 2,326 unconfirmed planetary candidates ("Kepler-22b" 2011). Despite the many early unfruitful efforts, exoplanet detection is now in full swing, with astronomers presently on the verge of discovering Earth-mass, terrestrial planets. A new study by Cassan et al. (2012) recently revealed that, on average, every star in our Milky Way may host one planet or more in an orbital range of 0.5–10 AU, implying that planets may be the rule rather than the exception in our Galaxy. Clearly many exciting exoplanet discoveries await. 1.2 The Radial-Velocity Method of Detecting Exoplanets Responsible for over 90% of currently detected exoplanets, the radial-velocity (RV) method has been by far the most successful technique for discovering exoplanets. The RV method uses the Doppler-shifted absorption lines in a host star's spectrum to 1 "The Extrasolar Planets Encyclopaedia." http://exoplanet.eu/ 5 determine the number of planets orbiting the star, along with the planetary masses and orbital distances. 1.2.1 Physical Detection Mechanism The mutual gravitational attraction of a host star and its planets introduces a relatively small, but detectable "wobble" in the host star's motion, as both the star and planets orbit around the entire system's center of mass, as depicted in Figure 1-1. As shown in Figure 1-2, this motion produces a characteristic RV-versus-time curve for the host star (a sine curve in the case of a circular orbit), with period, P, and velocity semiamplitude, K. Stellar RV measurements detect this reflexive orbital motion of the host star around the star-planet barycenter, which is a function of the mass of the star, the mass of the planet, and the separation of the two objects. Combining and manipulating Kepler's 3rd Law, the center of mass relationship, and the orbit equation for an ellipse produces the equation for the host star's velocity semi-amplitude: 1 2G 3 m p sin i Ks 2 P ms m p 3 1 1 e2 , (1.1) where Ks is the host star's velocity semi-amplitude, ms is the mass of the star, e is the orbit's eccentricity, i is the inclination angle (defined such that the orbital plane is perpendicular to the plane of the sky when i = 90°, see Figure 1-3), and mp is the mass of the planet. Generally speaking, since ms >> mp, Ks is proportional to mpsini. Therefore, measuring the host star's RV only provides a minimum mass of the exoplanet unless the orbital inclination can be determined through independent means, such as astrometry or a 6 planetary transit of the host star. Since orbital inclination is random, the probability that an orbit has an inclination in the range Δi = i1 – i2 is given by P(Δ) = cos i1 – cos i2. This means that roughly 90% of all exoplanetary systems can be expected to have an inclination greater than 30°, producing measured RV amplitudes that are within a factor of two of the true stellar RV (Lunine 2008). On average, the true mass of the exoplanet will be 1.57 times greater than that derived through observation (Casoli and Encrenaz 2007). Figure 1-1: An exoplanet and its host star orbiting their common center of mass, denoted by the red plus sign (figure not to scale). 7 Figure 1-2: Sample radial-velocity curve for a star with an orbiting exoplanet with period P. The exoplanet induces radial-velocity semi-amplitude, K, in the host star's motion (Clubb 2008). Figure 1-3: Definition of orbital inclination. When i = 90° the orbit is seen edge-on, and when i = 0° the orbit is seen face-on (Casoli and Encrenaz 2007). The RV method has intrinsic biases that lead to selection effects based on characteristics of the detected exoplanets. It is clear from Equation 1.1 that planetary systems with a short period (hence, a small orbital semi-major axis) and high planetary mass will produce the largest stellar reflex velocity. Table 1-1 provides a comparison of 8 the induced RVs on a solar-mass star by a few of our Solar System planets at various orbital distances. For example, a Jupiter-mass planet located at 1 AU from a solar-mass star produces a 28.4 m/s RV, while an Earth-mass planet at 1 AU only produces a 9 cm/s RV. Therefore, it comes as no surprise that many of the early exoplanet discoveries using the RV method have been "hot Jupiters." Figure 1-4 displays a plot of exoplanet minimum mass versus semi-major axis for planets detected using the RV method. There is a distinct lack of planets at low masses and large orbits, since Equation 1.1 clearly demonstrates the RV method's preference for detecting high-mass, short-period planets. The red line in the plot denotes the Earth's mass and, as of yet, no Earth-mass exoplanets have been found (no planets are located below the red line in the plot). The RV method is also inadvertently biased towards highinclination orbits, due to the stellar RV semi-amplitude's mpsini dependence. It should be noted that the RV method can be used to determine the planetary mass, orbital semi-major axis, eccentricity, and period, but offers no insight into the size of the planet. The transit method, by which a photometric dip in light can be detected as an exoplanet crosses the face of its host star, provides radius and inclination angle measurements for an exoplanet, but no mass measurement. Combined, the RV and transit methods can provide a planetary density based on the planet's mass and radius. 9 Table 1-1: Radial velocity comparison table for various planets orbiting a solar-mass star (Pasquini 2009). Planet Mass Jupiter Jupiter Neptune Neptune Super-Earth (5 M ) Super-Earth (5 M ) Earth Orbital Distance (AU) 1 5 0.1 1 0.1 1 1 Radial Velocity 28.4 m/s 12.7 m/s 4.8 m/s 1.5 m/s 1.4 m/s 0.45 m/s 9 cm/s Figure 1-4: Exoplanet minimum mass v. semi-major axis for planets detected using the RV method (as of March 2012). The mass of Earth is highlighted by the red line and the green circle is located at Earth's orbital distance from the Sun.2 2 exoplanets.org 10 1.2.2 Spectral Information The stellar RV induced by a planetary companion causes the host star's spectrum to be redshifted and blueshifted due to the Doppler effect as the star moves away from and towards an observer, respectively, as shown in Figure 1-5. This Doppler shift can be measured by detecting the minute position changes of stellar absorption lines using a spectrograph. Figure 1-6 displays the solar spectrum with many dark absorption lines. Since our atmosphere absorbs most of the ultraviolet (UV) and mid-infrared (mid-IR) portions of the electromagnetic spectrum, ground-based RV detection relies on spectral lines in the visible and near-IR (NIR) regions. The absorption line position changes are indeed miniscule; for a typical spectrograph with a resolving power (λ/Δλ) of R ~ 50,000 in the visible/NIR, an RV amplitude of 1 m/s corresponds to a shift of approximately 0.001 pixels on the detector. Figure 1-5: Radial velocity Doppler shift in a host star's spectrum ("The Radial Velocity Method" 2007). 11 Figure 1-6: Solar spectrum with absorption lines (Pogge 2006). Radial-velocity precision depends on three main factors: the signal-to-noise (SNR) ratio across the continuum and both the depth and width of the spectral absorption lines (Lovis and Fischer 2010). Assuming Gaussian shapes for spectral lines, the error in RV precision can be approximated by: RV FWHM , C SNR (1.2) where FWHM is the full width at half maximum of the spectral line, C is the contrast (the line's depth divided by the continuum level), and SNR is the signal-to-noise ratio in the continuum (Lovis and Fischer 2010). Equation 1.2 shows that narrow, deep absorption lines (small FWHM and large contrast) introduce less RV error than broad, shallow absorption lines, since it is essential to find an absorption line's centroid in order to determine its relative RV shift. 12 However, since the absorption line shifts are so small, hundreds of stellar spectral lines, in addition to wavelength calibration spectra, must be used to achieve high precision. Consequently, the density of stellar absorption lines in the spectral region of interest plays a major role in RV exoplanet detection. Since each spectral line contains an individual measurement of the stellar RV, if a total of M spectral lines are used for the measurement, then the error will be reduced by a factor of M over a single line measurement (Hatzes et al. 2010). According to Poisson statistics, the SNR also scales proportionally with N , with N being the number of photons collected, so longer exposure times will also, in principle, increase RV precision. The spectrograph's resolution plays a role in determining RV precision as well. Low spectral resolution will cause all of the individual lines to blur together and much RV information will be lost. A spectrograph with a large resolution must be used so that the lines can be separated and their wavelength centroids calculated. A higher spectral resolution also produces a higher velocity resolution per pixel. Nevertheless, it should be noted that higher spectral resolution runs counter to higher SNR and greater wavelength coverage. Higher-resolution spectrographs often have limited wavelength coverage, which reduces the number of available spectral lines, and they spread the light over more detector pixels, decreasing the SNR per pixel for a given exposure time (Hatzes et al. 2010). 13 1.2.3 Stellar Noise Beyond the technical spectrograph complications of measuring precise stellar radial velocities, the stars themselves can introduce measurement limitations. This socalled "stellar noise" can be introduced by several sources, including p-mode oscillations, granulation, magnetic activity, and rotation. P-mode oscillations are pressure modes that are stochastically excited at the outer surface of stars through turbulent convection and cause jitter in RV measurements. These oscillations have periods of a few minutes in solar-mass stars and produce typical RV amplitudes of a few centimeters per second for each mode (Bouchy and Carrier 2001; Kjeldsen et al. 2005). When a large number of modes are combined through superposition, the total RV variation can be several meters per second. Longer exposure times can be used to average over these variations. Granulation is the pattern of bright upflows and dark downflows due to largescale convective motion in the outer layers of stars with a convective envelope. For the Sun, these motions are generally 1–2 km/s in the vertical direction for a typical timescale of 10 minutes, though the large number of granules (~106) averages out this large velocity field (Lovis and Fischer 2010). Jitter due to granulation is expected to be at the meters-per-second level for the sun and represents a significant noise source for submeter RV precision. Stellar magnetic activity leads to a number of phenomena that impact RV measurements. Magnetic fields spawned by convective dynamos result in stellar flares with kilometer-per-second outflows, introducing unwanted blue shifts in spectral line 14 profiles (Fischer 2010). Stellar activity is tied to the stellar rotation rate, which produces a projected rotational velocity of vsini measured in kilometers per second, again causing jitter in RV measurements (Noyes and Weiss 1984). The RV precision is approximately inversely proportional to rotational velocity, scaled to a nominal slowly rotating star with an equatorial velocity of 2 km/s (Hatzes et al. 2010). Stars with higher rotational velocities have degraded RV precision, since the rotation broadens and reduces the depth of the spectral features (Bouchy et al. 2001). 1.3 Exoplanet Habitability The logical addendum to the question of whether there exist worlds beyond our own is that, if they do exist, are they habitable? Though many philosophers and scientists throughout human history have pondered this question, our current technology now offers the exciting prospect of actually detecting habitable exoplanets. We obviously know that our Universe can support life – we are here! Nevertheless, the discovery of extraterrestrial life beyond our Solar System would have profound philosophical implications, providing evidence for evolutionary theories of planets and life, and even offering clues into the past evolution of Earth and possible anthropogenic effects (e.g., the Gaia hypothesis, that life itself has played a major role in keeping Earth habitable). Defining η , the fraction of stars with a potentially habitable planet, is an important consideration in the search for exoplanets and would offer insights into the habitability of the Milky Way and the Universe in general. 15 The study of Earth's atmosphere and climate has long been of interest to geologists, biologists, and pre-biotic chemists, but now astronomers hope to partner with these researchers to learn more about the conditions in which life could develop beyond Earth (Kasting and Catling 2003). Yet we must fundamentally understand that tracing the detailed history of life on Earth does not provide all of the answers, since we are but one example of what the "branching ratio" might be for the development of life (Tarter 2007). Life on Earth has an amazing environmental adaptability, surviving in a pH range from <1 to 12, pressure from <1 to >500 bars, salinity from zero to saturation, and temperature from >100° C to below –40° C (Tarter et al. 2007). Though some of these conditions only allow organisms that are true extremophiles, a few common requirements exist for life as we know it. Terrestrial life requires liquid water and land is required to create stable temperature and pressure conditions (Kasting 2011). In particular, we can use the presence of water as an indicator of potential life on exoplanets. Despite the debate from biochemists that liquid water might not be required for life, it may be a de facto requirement in the search for extraterrestrial life, because we do not currently know what spectroscopic biosignatures to look for without it (Kasting 2011; Zugger et al. 2011). At the very least, using water as an indicator of potential life will provide us with a subset of life-bearing planets (Scalo et al. 2007). After all, we know from personal experience that liquid water can support life. 16 1.3.1 The Habitable Zone The concept of a circumstellar habitable zone (HZ) dates back to the 1950s and the research of Shapley (1953), Strughold (1953), and Huang (1959). Modern definitions, such as that of Kasting et al. (1993), define the habitable zone as the orbital distances from a host star over which liquid water can exist on a planet's surface. Figure 1-7 provides a visual representation of the HZ for different stellar spectral classifications. Clearly, the size and distance of the HZ from the host star depends on spectral type, further depicted in Figure 1-8. The size of the HZ depends on the luminosity, mass, and activity of the host star. A rough estimate for the location of the HZ based on the host star's mass is given by: a HZ M 3 2 , (1.3) where a is the orbital distance in AU and M is the host star's mass in solar units (Scalo et al. 2007). Earth's HZ extends from about 0.95 AU (the water-loss limit) to an outer edge between 1.37 AU and 2.4 AU depending on the amount of warming due to atmospheric feedback mechanisms (Kasting and Catling 2003). The HZ is not static for a host star; stellar evolution acts to evolve the HZ as well. According to solar evolutionary models (Newman and Rood 1977; Gough 1981; Gilliland 1989), the Sun was approximately 30% dimmer when the Solar System formed ~ 4.6 Gyr ago and will increase to 3 times its present luminosity by the time it evolves off the main sequence in 5 Gyr. 17 Figure 1-7: The zero-age main sequence habitable zone. Stellar classification is provided on the vertical axis and the dotted line indicates the tidal locking radius after 4.5 Gyr (Kasting et al. 2003). Figure 1-8: Schematic of the habitable zone (green zone) of different host stars (not to scale). "Hotter" stars roughly translates to O, B, and A stars, "Sunlike" stars to F and G stars, and "Cooler" stars to K and M stars ("Kepler Artwork"). 18 The evolution of the HZ leads to a continuous habitable zone (CHZ), first introduced by Hart (1979). The CHZ indicates the orbital distances over which liquid water can exist for an extended period of time on a planet independent of stellar evolutionary changes. A commonly accepted value for this time period is τhab ~ 4 Gyr, the time needed on Earth for intelligent life to emerge (Buccino et al. 2006) Being located in the HZ or even in the CHZ is no guarantee for life on a planet, since the evolution and survival of life also depends on environmental factors such as asteroid impacts, as well as planetary size, atmosphere, geology, and rotation. Furthermore, the tidal forces between a host star and a planet can significantly change the planet's orbit, and hence the habitable lifetime (Barnes et al. 2008). It therefore can be difficult to infer a planet's past or future habitability based on its present location. Tides decrease a planet's semi-major axis and reduce its orbital eccentricity until the orbit is circularized, which may leave the planet past the inner edge of the HZ. For planets to remain in the CHZ long enough for life to develop, they must form with a relatively low eccentricity, a fact that is also dictated by the circular symmetry of the HZ (Barnes et al. 2008). The circumstellar HZ is not the only location capable of supporting liquid water. Jupiter's moon Europa is completely enveloped in a thick ice sheet, which is suspected to cover a subsurface liquid water ocean heated by friction from Jupiter's tidal forces (Greenberg 2005; Blankenship et al. 2011; Solomonidou et al. 2011). The circumstellar HZ inherently relies on the star as the primary source of life-supporting energy, but Europa could possibly support life at hydrothermal vents beneath its ocean, similar to life 19 already known to exist in the darkest depths of Earth's oceans. However, since the potential life on Europa and a few of our other Solar System moons can only be detected in situ, having no noticeable effect on the moons' atmospheres, we must rely on the circumstellar HZ definition to guide our current searches for life in exoplanetary systems. 1.3.2 Exoplanet Atmospheres Planetary size is an important consideration for habitability, since a small planet will lack the necessary gravity to maintain a life-supporting atmosphere, while a large planet will accrete hydrogen and helium to the point of becoming a gas giant. Ten Earth masses (10 M ) is thought to be the upper limit for a rocky planet to avoid transitioning from terrestrial planet to gas giant. A planetary atmosphere is also a critical indicator of biosignatures, since we must currently rely on the indirect atmospheric effects of extraterrestrial life for detection. One of the most dramatic impacts of a planet's atmosphere on habitability is its ability to affect the planet's surface temperature. Treating a planet as a blackbody with effective temperature Te, the planetary energy balance is given by: Te 4 S 1 A , 4 (1.4) where σ is the Stefan-Boltzmann constant (= 5.67 × 10-8 W/m2/K4), S is the stellar flux at the planet's orbit, and A is the planetary albedo. Solving for the effective temperature of Earth (S = 1370 W/m2, A = 0.3) yields Te = 255 K. This is obviously much colder than typical Earth temperatures. In reality, Earth is not a blackbody, and instead, it is warmed 20 by our atmosphere through the greenhouse effect. Infrared radiation emitted by the Earth's surface is absorbed in the atmosphere by “greenhouse gases,” such as CO2 and water vapor, and re-emitted back towards the Earth. This raises the global average surface temperature Ts to ~288 K. The magnitude of the greenhouse effect is the difference between Te and Ts: ΔTg = Ts – Te = 33 K. (1.5) For Earth, water vapor provides about two-thirds of this warming, while CO2 contributes most of the remaining third (Kasting and Catling 2003). Climate feedback processes, such as the carbonate-silicate cycle, can help maintain the necessary levels of greenhouse gases in a planetary atmosphere to sustain life. Over a time scale of approximately half a million years, the level of CO2 in Earth's atmosphere is controlled by slow interactions with Earth's surface. CO2 dissolved in rainwater will form carbonic acid and eventually weather calcium and magnesium silicates in rocks. Streams and rivers transport the products of silicate weathering to the ocean, where organisms use them to make shells of calcium carbonate (CaCO3). Most of the shells re-dissolve when the organisms die, but a fraction of them fall to the ocean floor and are buried in sediments. Walker (1991) notes that this portion of the carbonatesilicate cycle does not require life. Although most CaCO3 is precipitated biotically today, the same reaction would occur in the absence of biota, but with a higher bicarbonate concentration. If the process ended here, all of the Earth's CO2 would end up in the crust and the planet would become uninhabitable. Fortunately, carbon reenters the atmosphere when the seafloor is subducted by the continental plates, exposing the carbon sediments 21 to high temperatures and pressures. As a result, the carbonates are recombined into silicates and CO2, which is vented back to the atmosphere by volcanoes. Weathering reactions, such as those occurring in the first step of the carbonatesilicate process, only proceed at reasonable rates in the presence of liquid water. If Earth cooled to a point that the oceans started freezing, silicate weathering would come to an end, and CO2 would begin to build up in the atmosphere (Kasting et al. 1993). This would lead to a higher greenhouse effect that would melt the ice. In the opposite case, exhibited by Venus in our Solar System, initial high temperatures will lead to water loss and the cessation of silicate weathering. Volcanic CO2 accumulates in the atmosphere unabated and a runaway greenhouse effect takes hold (Kasting and Catling 2003). A substantial planetary mass (≥0.3 M ) is needed to drive the plate tectonics necessary for the carbon-silicate cycle (Raymond et al. 2007). Undoubtedly, the carbonate-silicate is significant in providing a negative feedback on temperature regulation and constrains the inner and outer edges of the HZ. A planet at the inner edge of the HZ will develop a dense, H2O-rich atmosphere that will begin to photodissociate, releasing hydrogen into space and eventually leading to loss of surface water and habitability. At the outer edge, the dense, CO2 atmosphere itself begins to condense and the negative feedback of the carbon-silicate cycle is lost (Lunine et al. 2008). 22 1.4 M Dwarfs as Exoplanet Host Stars M dwarfs offer a very unique proposition for habitable exoplanets, presenting a radiation environment that is much different than that from our own Sun. Discovering life on an M dwarf exoplanet would be a true testament to the versatility of life. About 45% of M dwarfs exhibit an IR excess in their spectra, indicative of a circumstellar disk necessary to form planets, suggesting that M dwarfs may frequently host exoplanets (Luhman et al. 2005). Though over 30 exoplanets have been found orbiting M dwarfs, they are mostly Super-Earths (~2–10 M ), so the drive continues to detect a true Earth analog (Mayor et al. 2009a; Engle et al. 2009). 1.4.1 Properties of M Dwarfs M dwarfs are by far the most populous stellar spectral classification, constituting about 75% of the stellar population, as shown in Figure 1-9. They are the least massive of stars, from 0.08–0.60 solar masses (M ) (Reid and Hawley 2006). Objects below about 0.08 M become degenerate before igniting hydrogen fusion and remain brown dwarfs. This factor of 7 range in mass is the fundamental feature influencing the basic properties of M dwarfs and the habitability of their planets. This is a larger mass interval than stellar spectral types A, F, G, and K combined and, given their sheer numbers, the mass of M dwarfs dominates the Milky Way (Scalo et al. 2007). 23 Figure 1-9: Population of known stars within 10 parsecs (~32.6 light years), binned by spectral type.3 The large range of relatively small M dwarf masses has a direct influence on their large range of small-valued luminosities. As displayed in Table 1-2, luminosity spans a range of almost 500 from early-type (M0) to late-type (M9) M dwarfs. Due to their lower temperatures and higher opacity, M dwarfs move luminosity outward through convection rather than radiation throughout their bulk. This convective mixing enables much more of the star to be available as nuclear fuel, resulting in main-sequence lifetimes of 50 Gyr for early-type M dwarfs to several trillion years for late-type M dwarfs (Tarter et al. 2007). Figure 1-10 shows the evolutionary track for several low-mass M dwarfs, ranging from about M4 to M9. Convective mixing also means that the core composition rises in mean molecular weight very slowly and so the emitted flux stays stable for 3 RECONS http://www.recons.org/ 24 billions of years, resulting in a slower evolution of the HZ and a larger CHZ. This is in contrast to solar-type stars, which undergo significant brightening during their mainsequence lifetimes. For example, though our Sun has increased in luminosity by 30% during its ~4.6 Gyr lifetime, a mid-M dwarf (M4–M5) would only increase in luminosity by <1% over the same time period (Engle and Guinan 2011). Figure 1-11 illustrates the extremely slow evolution of M dwarfs compared to other stellar spectral types. In fact, given the age of the Universe (~13.7 Gyr), no M dwarfs have even come close to completing their final evolutionary stages. Table 1-2: Physical characteristics of each M dwarf subclass (Kaltenegger and Traub 2009). Spectral Type T(K) R (Rsun) Mass (Msun) L/100 (Lsun) Mv (mag) M0 3800 0.62 0.60 7.2 9.34 M1 3600 0.49 0.49 3.5 9.65 M2 3400 0.44 0.44 2.3 10.12 M3 3250 0.39 0.36 1.5 11.15 M4 3100 0.26 0.20 0.55 12.13 M5 2800 0.20 0.14 0.22 16.0 M6 2600 0.15 0.10 0.09 16.6 M7 2500 0.12 ~0.09 0.05 18.8 M8 2400 0.11 ~0.08 0.03 19.8 M9 2300 0.08 ~0.075 0.015 21.4 25 Figure 1-10: Evolution of several low-mass M dwarfs on the Hertzsprung-Russell diagram. The inset diagram displays the main-sequence lifetimes as a function of mass (Laughlin et al. 1997). 26 Figure 1-11: Comparison of stellar spectral evolution (Engle and Guinan 2011). Since planetary climate and atmospheric chemistry are significantly dependent on the flux from the host star, the low luminosities of M dwarfs will be the driving factor limiting the habitability of their exoplanets. Table 1-3 details the distance from the host star at which a planet would receive the same amount of flux as Earth does from our Sun for different M dwarf subclasses. The M dwarf HZ certainly lies much closer to the host star than it does for a solar-type star. Table 1-3: Orbital distance of equivalent solar flux for M dwarfs (solar flux = 1360 W/m2) (Heath et al. 1998). Spectral Type Orbital radius (AU) Orbital radius (stellar radii) G2 M0 M1 M2 M3 M4 M5 M6 M7 M8 1.00 0.26 0.23 0.19 0.15 0.12 0.10 0.07 0.04 0.02 214 56 48 41 33 27 21 14 9 5 27 The low luminosity of M dwarfs is directly coupled to their low temperatures, which cause M dwarf spectra to peak in the NIR. Figure 1-12 illustrates that M dwarfs radiate most of their energy in the Y (0.9–1.1 μm), J (1.1–1.4 μm), and H (1.45–1.8 μm) spectral bands of the NIR. The low temperature of M dwarfs supports the formation of molecules such as TiO, VO, H2O, and CO (Rodler et al. 2011). TiO and VO absorb much of the blueward stellar flux, while H2O and CO create a forest of absorption lines in the NIR that are well-suited for RV measurements. Figure 1-12: Spectral energy distribution for an F, G, and M star (Kasting et al. 1993). 1.4.2 M Dwarf Exoplanets For years, astronomers ruled out M dwarf exoplanets as potential abodes for life. Two main reasons were put forward for excluding M dwarfs from the habitable planet 28 discussion: the flare activity of young M dwarfs and the tidal locking and synchronous rotation that results from planets orbiting close to their host star. 1.4.2.1 M Dwarf Activity The largest uncertainty for the habitability of M dwarf exoplanets is the shortterm variability of the radiation environment, especially for young M dwarfs, and the corresponding impact to a planetary atmospheric and climate system (Scalo et al. 2007). The rotation of a star and the motion of gas in its convective layer generate magnetic fields that produce activity phenomena, such as flares. The efficiency of this stellar magnetic dynamo is governed by two main parameters: the stellar rotation rate and the depth of the convective layer. Since the stellar convective envelope is much thicker for M dwarfs, at a given rotation period, M dwarfs are more active than solar-like G stars (Khodachenko et al. 2007). M dwarf flares can last from minutes to many hours and emit energies up to 1027 Joules, 102 to 107 times greater than typical solar flares (Tofflemire et al. 2012). Although the luminosity of M dwarfs is remarkably steady over their mainsequence lifetimes, flares lead to periodic transient increases in their X-Ray, UV, visible, and IR emission output, along with emission of charged particles as stellar wind. Young M dwarfs spin rapidly and therefore have strong magnetic dynamos and high levels of activity. As M dwarfs age, they lose angular momentum via magnetized stellar winds and their activity levels decrease (Engle and Guinan 2011). The activity lifetime and fraction of active M dwarfs increases significantly progressing from earlytype to late-type, as depicted in Figures 1-13 and 1-14, respectively. 29 Figure 1-13: Activity lifetime of M dwarfs (West et al. 2008). Figure 1-14: Fraction of active M dwarfs as a function of subclass (West et al. 2008). Since late-type M dwarfs remain in their initial state of high activity for a longer period of time, this could potentially pose problems for hosting habitable exoplanets. The lowest-mass M dwarfs also take longer to reach the main sequence (0.3–1 Gyr) after formation, during which their luminosity decreases by two orders of magnitude (Laughlin 30 et al. 1997). Since the location of the HZ directly depends on stellar flux, the HZ for the lowest-mass M dwarfs evolves significantly, as illustrated in Figure 1-15. Figure 1-15: Evolution of low-mass M dwarf habitable zone boundaries (Khodachenko et al. 2007). The high flare activity and stellar wind of M dwarfs in their early stages can also erode the atmosphere of exoplanets located in the HZ, since they orbit much closer to their host star. This is of particular concern for slowly-rotating planets, which lack the planetary dynamo and resulting magnetic field that protect the planet from stellar wind (Lunine et al. 2008). However, Lammer et al. (2008) concludes that if an Earth-mass planet orbiting within the HZ of a highly-active M dwarf can generate a strong enough intrinsic magnetic field, its atmosphere may survive erosion as long as the X-ray and extreme UV (XUV) fluxes are less than 50–70 times that of our present Sun. 31 1.4.2.2 Tidal Locking and Synchronous Rotation Tidal forces from a host star act to reduce a planet's rotation rate, which is linked to the planetary magnetic field that protects the planet from stellar activity. M dwarf exoplanets located in the HZ are tidally locked in synchronous rotation with their host star, meaning that one side of the planet permanently faces the host star. The tidal lock orbital radius for different spectral type host stars, indicated by the dotted line in Figure 1-7, predicts that Earth-like planets located in the HZ of M dwarfs will become tidally locked after 4.5 Gyr of evolution (Kasting et al. 1993). It was long assumed that the persistent lack of flux on the dark side of a tidally locked planet would lead to atmospheric collapse. In this scenario, the low temperatures on the far side of a synchronously rotating planet would cause the volatile compounds in the atmosphere and the oceans to freeze out to form a giant ice cap (Segura et al. 2005). However, as shown in simulations by Joshi et al. (1997) and Joshi (2003), atmospheric collapse can be avoided if longitudinal winds exist that transport heat from the dayside to the nightside. The higher the atmospheric mass, the larger the horizontal atmospheric motion (advection) will dominate over radiation and reduce the day/night temperature gradient. The key determination is that the advective time scale must be less than the time scale for thermal relaxation. Given the terrestrial value of solar flux, a CO2 and H2O atmosphere on a synchronously rotating exoplanet will only condense on the dark side when the surface pressure is approximately 30 mb (Joshi et al. 1997). This pressure is so low because as pressure decreases, the frost temperature also decreases, further staving off atmospheric collapse. Joshi et al. (1997) concludes that a minimum 32 surface pressure of 1000–1500 mb of CO2 is needed for liquid water to exist on the darkside of a tidally-locked M dwarf exoplanet located at the inner edge of the HZ. Heath et al. (1998) argues that a similar reduction in the day/night side temperature gradient can be achieved with heat flow through a sufficiently deep ocean. It should be noted that not all planets within a host star's tidal lock radius for a certain evolutionary time period will become synchronous rotators. Mercury is located within the tidal lock radius for our Sun, but is locked in a 3:2 spin/orbit resonance, meaning that Mercury rotates three times for every two orbits around our Sun. This would greatly reduce the chance of atmospheric collapse for M dwarf exoplanets in a similar scenario. 1.5 Current Challenges of Near-Infrared Precision Radial-Velocity Measurements To date, the vast majority of precision RV measurements have been made of solar-type stars in the visible, aided by the large number of deep, sharp absorption lines at wavelengths shorter than 0.7 µm. These visible RV programs have also benefited from the relative advancement of visible wavelength spectrographs compared to other spectral regions (Bean et al. 2010). The High Accuracy Radial velocity Planet Searcher (HARPS) at the European Southern Observatory (ESO) in Chile has already achieved sub-1-m/s precision in detecting exoplanets in the visible (Mayor et al. 2009b). However, the quest to discover exoplanets around M dwarfs requires a shift in focus away from the visible and into the NIR. The NIR offers a unique set of challenges for precision RV work, 33 namely a suitable calibration method, atmospheric telluric line spectral contamination, and increased modal noise in the optical fibers that couple the telescope to the spectrograph. 1.5.1 NIR Calibration Sources Spectrographs used in precision RV work are typically calibrated using known emission- or absorption-line spectra to compare to the Doppler-shifted stellar spectrum. The ideal comparison spectrum contains spectral lines that are of known wavelengths determined from atomic physics, individually unresolved (yet resolved from each other), uniformly spaced, cover the entire spectral region of interest, have uniform intensity, have long-term stability, and do not reduce the object spectrum's SNR (Murphy et al. 2007). Achieving such a stable, well-characterized reference is no small achievement, especially when transitioning from the visible to the NIR, where no heritage exists. I2 gas cells and Th–Ar emission lamps have already been used for precision RV measurements in the visible, but the relatively new concept of the laser frequency comb may ultimately hold the answer as the preferred calibration source for precision RV calibration. 1.5.1.1 I2 Gas Cell For the absorption cell method, a sealed glass cell of I2 gas is placed in the science beam path, adding a dense forest of absorption lines in the 500–620 nm regime of the stellar spectrum. A model of the observed stellar spectra with the superimposed forest of 34 absorption lines and the point spread function (PSF) of the spectrograph are used to obtain the RV shift of the stellar absorption lines (Butler et al. 1996; Martín et al. 2011). Although this technique has the advantages of high line density, stability, and the need for only one optical fiber to be fed into the spectrograph, the I2 gas cell has limited wavelength coverage outside of the visible. 1.5.1.2 Th–Ar Lamp A second popular method of RV calibration utilizes a Th–Ar hollow-cathode lamp. Usage of hollow-cathode lamps for RV calibration dates back to the early 1990s in the field of stellar seismology (Brown et al. 1991). The emission spectrum of the Th–Ar lamp is recorded simultaneously with the observed stellar spectrum by feeding two separate optical fibers into the spectrograph during the exposure. Instrumental shifts are minimized since the stellar and calibration spectra are recorded concurrently, appearing adjacent to each other on the detector. Th–Ar lamps can provide coverage from 250 nm to 4500 nm and offer the benefit of less computation, since the Th–Ar lines are not directly mixed with the stellar spectrum, as with the I2 gas cell. Despite the advantages, Th–Ar lamps suffer from aging effects, which can change the position of the emission lines or causes lines to appear or disappear over time (Hatzes et al. 2010). The use of two optical fibers necessitates higher mechanical stability of the spectrograph and temperature or pressure changes can also cause the calibration lines to shift. In particular, the argon emission lines are susceptible to this effect and cannot be 35 used in precision RV measurements (Lovis and Pepe 2007). The bright argon lines can also lead to scattered light and saturation effects if the lamp is run with high current. 1.5.1.3 Laser Frequency Comb The laser frequency comb (LFC) is an emerging technology that holds great promise as a reference for precision RV measurements. A Ti-doped sapphire or Er-fiber mode-locked, femtosecond-pulsed laser produces a series of bright, narrow (<1 MHz), uniformly-spaced (~0.25–1 GHz) laser modes at the pulse repetition rate, frep, which is governed by the adjustable laser cavity length (Braje et al. 2008; Murphy et al. 2007). These laser frequency modes can cover hundreds of nanometers in the visible and NIR due to non-linear optical fiber broadening (Osterman et al. 2011). The laser mode frequencies are directly referenced to atomic frequency standards such as the Global Positioning System (GPS), ensuring their stability and accuracy to better than 1 part in 1011 (Quinlan et al. 2010). This is sufficient for cm/s precision RV measurements, an order of magnitude improvement over traditional absorption cells and emission lamps. Directly linking the frequency standard to GPS offers the additional benefit of allowing comparison of measurements between different instruments and over long periods of time (Lawson et al. 2009). The intrinsic frequency spacing of the laser emission lines is too dense to be resolved by a typical grating spectrograph, but a Fabry-Pérot etalon can be used to reduce the line density. This is achieved by setting the etalon's free spectral range (FSR) to an integer multiple of frep (FSR = m × frep, with m ~ 10–100) (Li et al. 2010; Steinmetz et al. 36 2008; Braje et al. 2008). This lets the etalon act as a mode filter by passing every mth comb line, allowing the LFC to match the spectrograph's resolution. Although LFCs represent the future of precision RV calibration and already have been used in preliminary tests on stars (Ycas et al. 2012), they are not readily available and are more expensive than other references. Nevertheless, LFCs represent a breakthrough in precision RV measurements, one that will soon prove particularly useful in the NIR detection of M dwarf exoplanets. 1.5.2 Telluric Absorption Unlike in the visible, NIR ground-based astronomical observations are hindered by the presence of strong H2O, O2, and CO2 telluric absorption lines, as seen in Figure 116. The depth of these telluric lines fluctuates on timescales of hours and is particularly dependent on the changing column density of water vapor and the exact pointing of the telescope (Lunine et al. 2008). Contamination by telluric lines introduces variability into stellar spectra, reducing the achievable RV precision. Telluric lines can be removed by using terrestrial absorption forward modeling algorithms (Gettel et al. 2011). 37 Figure 1-16: Atmospheric transmission in the NIR (Geballe 2010). 1.5.3 Optical Fibers Fiber optics made their debut in astronomical telescope applications in the late 1970s due to significant improvements in that technology made by the communications industry (Barden 1994). Today, multimode optical fibers play a critical role in precision RV measurements, coupling telescope and calibration light to the spectrograph entrance slit. This highlights the main advantage of optical fibers in this application, namely, their ability to feed a spectrograph that is physically removed from the telescope in a temperature- and pressure-controlled room. As noted earlier, mechanical, temperature, and pressure stability are essential in accurately measuring precise RVs. Multimode fibers are used over single mode fibers since they are easier to couple light into and their larger cores allow more light into the spectrograph (Chen et al. 2006; Bland-Hawthorn and Kern 2009). However, unlike their single mode counterparts, multimode fibers 38 propagate many modes along their length that cause unpredictable changes in the spectrograph's instrument profile (IP). 1.5.3.1 Modal Description Solving Maxwell's equations for an optical fiber yields a discrete set of propagation solutions, known as modes, which depend on the wavelength of the propagating radiation, fiber refractive index profile, and the fiber dimensions (Haynes et al. 2004). A simple ray model is useful in conceptualizing modal propagation through optical fibers. Figure 1-17 displays a geometrical representation of modal propagation through a multimode optical fiber, with the fundamental mode entering the fiber input face on-axis and higher-order modes entering at increasing angles up to the fiber's numerical aperture (NA). Coherent light entering a fiber within a particular angular range (i.e., light with a defined input focal ratio) will excite a unique set of modes and produce a distinctive modal power distribution (MPD) at the fiber output. The MPD presents itself as a speckle pattern in the fiber's far-field, due to the interference of the electric-field patterns of the individual fiber modes (Wood 1984). A change in the input illumination of a fiber alters the MPD and, therefore, the far-field speckle pattern. Light entering a perfect fiber with no irregularities or variations will exit with the same modal distribution and focal ratio as the input. However, non-uniformities in a fiber, such as index of refraction variations, fiber diameter variations, core-cladding interface irregularities, and core scattering centers affect modal propagation in a quasirandom manner, redistributing power between the various modes. In addition to these 39 internal factors, external stresses, temperature fluctuations, and beam truncation also change the modal distribution in a fiber. An example of these effects is shown in Figure 1-18, using the ray model for modal propagation. Light scatters into any given mode from many other modes, so that the light exhibits a quasi-random phase and amplitude variation in the resulting far-field speckle pattern (Corbett and Allington-Smith 2006). Due to its stochastic nature, this modal diffusion has been successfully modeled as a Gaussian process, i.e., the probability of neighboring modes exchanging power follows Gaussian statistics (Haynes et al. 2011). Figure 1-17: Mode propagation through an optical fiber (Corbett and Allington-Smith 2006). Even if light is initially only coupled into lower-order modes, higher-order modes inevitably will be excited through the redistribution of modal power due to the fiber irregularities and external factors. Fibers, therefore, have the tendency to decrease the focal ratio of input light, i.e., light will exit a fiber within a larger angular cone than when it entered the fiber, an effect known as focal ratio degradation (FRD). FRD worsens as a fiber is fed with slower-input focal ratios and has the effect of reducing a spectrograph's throughout efficiency and resolution. 40 Figure 1-18: Effect of fiber non-uniformity on modal propagation (Corbett and Allington-Smith 2006). 1.5.3.2 Modal Noise and Scrambling The number of excited modes in a uniformly-illuminated fiber is given by: 1 d NA M , 2 2 (1.6) where d is the fiber core diameter, λ is the light's wavelength, and NA is the fiber's numerical aperture (i.e., maximum coupling angle). Figure 1-19 displays the number of modes per wavelength in the NIR for an optical fiber with a core diameter of 200 µm and NA = 0.22. The number of propagating modes falls as λ–2, so there are fewer modes in the NIR relative to the visible. The relationship between the number of modes in a fiber and the corresponding SNR ratio is demonstrated by: SNR 2 v M 1 , 1 2 (1.7) where the vignetting factor is given by ρ2= Ad / Af,; Af is the illuminated area of the fiber end face and Ad the effective detector area illuminated by the fiber; and v is the visibility factor, which is a measure of the input light's coherence (i.e., the visibility increases with 41 higher spectral resolving power R, which reduces the SNR). Substituting Equation 1.6 into Equation 1.7, it is clear that SNR decreases inversely proportional with wavelength. Consequently, the lower number of modes in the NIR can add additional noise to PRV measurements if not mitigated in some manner. Figure 1-19: Number of modes in the far-visible and NIR for a multimode optical fiber (200-µm-diameter core and 0.22 NA). Modal noise, the time variation of the MPD between successive exoposures, is of particular concern when measuring precise RVs, since variation of the MPD will change the instrument profile (IP). Rawson et al. (1980) noted three conditions necessary for modal noise: a sufficiently narrow source spectrum, spatial filtering at the output plane, and either a source wavelength shift, fiber movement, or both. Astronomical RV spectroscopy meets all three of these conditions, since each pixel in a high-resolution spectrograph's detector subtends a very narrow spectral width, the spectral slit inevitably 42 vignettes the spectrograph's input fiber, and the fiber link moves with the telescope (Baudrand et al. 1998). To elaborate on the first criterion, modal noise is only a major concern for coherent light. When observing light of a broad bandwidth, the speckle contrast is dramatically reduced due to the much larger number of excited modes (Lemke et al. 2011). However, for very high-resolution spectrographs, the light at the detector becomes effectively coherent to the point that the SNR is no longer photon-limited, but modal noise-limited (see Equation 1.7) (Lemke et al. 2010). The stability of the spectrograph's IP is crucial in measuring the sub-pixel Doppler shift of stellar absorption lines induced by exoplanets on their host stars. Manifested as subtle temporal variations of the wavelength-dependent IP in science, calibration, or flat field exposures, modal noise produces increased noise in extracted stellar spectra. In addition to the fiber irregularities and external stresses, modal noise can also be caused by telescope guiding errors, seeing variations, and optical misalignments, which can all shift the point spread function (PSF) on the detector (Hunter and Ramsey 1992; Boisse et al. 2011). Noisy stellar spectra lead to errors in measuring stellar radial velocities, which affect the inferred number of exoplanets, along with their masses and orbital distances (see §1.2.2 – Spectral Information). Optical aberrations in instrument optics and fiber imperfections can cause RV shifts greater than 1 m/s (Lo Curto et al. 2010). It is important to note that modal noise cannot be removed through flat fielding, since the output speckle pattern changes with small perturbations, which inevitably occur when replacing a reference source with the telescope (Chen et al. 2006). 43 Optical fibers do provide better illumination stability for a spectrograph than a simple aperture or slit because fibers smooth out seeing and guiding variations due to their modal scrambling characteristics (Barden 1998). Fibers are particularly good at scrambling light azimuthally, though radial scrambling is not complete and retains some information about the fiber's input illumination (Hunter and Ramsey 1992; Avila et al. 2006). The fraction of light at the fiber output that retains information of the input's radial position increases as the fiber is fed with faster input focal ratios, so a tradeoff often must be made between limiting FRD and achieving sufficient fiber scrambling. Scrambling gain, the ratio of the displacement of a point source illuminating a fiber to the corresponding PSF shift on the detector, is an important measurement to quantify the amount scrambling within a fiber (Avila et al. 2006). Figure 1-20 illustrates this concept, represented in equation form by: G d f D. (1.8) F Mechanical agitators and optical scramblers can be used to increase fiber scrambling ability and reduce any dependency on variable illumination conditions. Baudrand and Walker (2001) were among the first to report that vibrating a multimode fiber near the output at a rate much faster than the exposure time averaged the power in the modes, creating a "scrambled" speckle pattern. Scrambling gains for this method of mechanical vibration are typically in the hundreds. Other types of mechanical scramblers mix fiber modes by submitting the fiber to systematic stresses, such as pressure, bending, or twisting (Avila et al. 2006). Hunter and Ramsey (1992) introduced the double 44 scrambler, in which a junction is made in the fiber coupling the telescope to the spectrograph. A hemispherical lens is then attached to each fiber face created by the junction, which has the effect of exchanging the angular and position dependency of light exiting and entering each fiber, as shown in Figure 1-21. This method improves fiber scrambling, but suffers from a reduction in throughput. Figure 1-20: The position shift of a point source at the fiber input corresponds to a linear PSF shift on the detector (Avila and Singh 2008). 45 Figure 1-21: Operation of a double scrambler. The top panel (a) shows that light exiting the output end of the input fiber at a particular position enters the input face of the output fiber at a particular angle. The lower panel (b) shows that light exiting the output end of the input fiber at a particular angle enters the input end of the output fiber at a particular position (Hunter and Ramsey 1992). 1.6 Current NIR Instruments Despite the many challenges listed in the previous section, on-sky RV measurements are steadily improving with several current NIR spectrographs. Martín et al. (2006) surpassed the precision achieved by previous optical RV work on late-type stars using NIRSpec (Near-InfraRed Spectrograph) on Keck II (R = 25,000; simultaneous wavelength coverage of 400 nm) to achieve 360 m/s RMS precision on RV measurements of the nearby brown dwarf LP944-20. Blake et al. (2007) used CO bands 46 and telluric features to reach a precision of 300 m/s on L dwarfs. CRIRES (CRyogenic high-resolution InfraRed echelle Spectrograph) on the Very Large Telescope (VLT) at ESO has a spectral resolving power R ≥ 100,000 in the 1–5 μm range and has achieved 5 m/s precision on late-type M dwarfs in the K band using an ammonia gas cell (Bean et al. 2010; Figueira et al. 2010). 1.7 Conclusions M dwarfs are prime targets for high-resolution, NIR spectrographs to detect HZ, Earth-mass exoplanets. Although the planetary environment offered by M dwarfs is certainly different than our Sun's, life has proven its amazing vitality here on Earth and there is no reason to doubt life's adaptability elsewhere. Since M dwarfs are by far the most numerous stellar classification, determining the habitability of M dwarf exoplanets would provide a lower limit on the habitability of our Milky Way. As with any new realm of scientific discovery, challenges must be overcome in transitioning precision RV measurements from the visible to the NIR. This thesis focuses on measuring and mitigating the challenge of modal noise in the NIR, which has not been extensively studied to date. 47 Chapter 2 Penn State Pathfinder Due to the low mass ratio between an M dwarf host star and an Earth-mass exoplanet, precision NIR RV measurements of M dwarfs offer the earliest opportunity to discover terrestrial exoplanets. Both the NASA/NSF commissioned Exoplanet Task Force Report (Lunine et al. 2008) and the Exoplanet Community Report (Lawson et al. 2009), published by blue-ribbon panels of exoplanet science and technology experts, prioritize extending RV precision to 1 m/s in the NIR and "fast-tracking" ground-based searches around M dwarfs towards finding Earth-mass exoplanets. The Penn State Astronomy and Astrophysics Department is involved in this effort with Pathfinder, a prototype spectrograph designed to explore the technical and data reduction issues that must be resolved in order to measure precise RVs of M dwarfs in the NIR (0.9–1.7 μm) from ground-based observations at the Hobby-Eberly Telescope (HET). Pathfinder derives much of its heritage from the Gemini-commissioned Precision Radial Velocity Spectrograph (PRVS) instrument study, in which Penn State was a partner (Jones et al. 2008). Pathfinder has already demonstrated <10 m/s RV precision in the NIR on solar observations using the Earth's rotation signal, but much work remains in order to achieve meter-per-second precision on M dwarfs (Ramsey et al. 2008). The lessons learned from the Pathfinder prototype and the research presented in this thesis will be used in the Habitable Zone Planet Finder (HPF), which is a future facility-class, high-resolution NIR spectrograph that will be capable of high-precision (<5 m/s) RV 48 measurements of Earth-mass planets orbiting mid- to late-type M dwarfs. For a detailed description of HPF's proposed design, please see Mahadevan et al. (2010). This chapter describes the optical design of Pathfinder and the on-sky performance of Pathfinder during engineering test runs at the HET. 2.1 Spectrograph Pathfinder is a fiber-fed, cross-dispersed echelle spectrograph (see schematic in Figure 2-1) that covers the NIR Y-, J-, and H-bands (0.9–1.7 µm) at a resolving power (λ/Δλ) of 50,000. Located in Davey Laboratory on the Penn State – University Park campus, the spectrograph is constructed primarily with standard optical mounting hardware on an 8 × 2 foot optical bench, but the fiber input and the detector are mounted on custom-made units. The optical bench is situated in a controlled-access interior lab to control temperature fluctuations that will influence RV measurements. Lakeshore PT102 platinum resistance thermometers with a Lakeshore 218 monitor indicate ambient room temperature variations are typically 0.1 K over several hours, while spectrograph temperature variations are on the order of 0.05 K. The spectrograph is fed by a single fiber optic cable whose input end is located on the roof of Davey Lab in one of the rooftop observatories. Over the past few years of testing, the instrument has been upgraded to improve efficiency. The input fiber coupling was changed to match the f/6 collimator with the f/4.2 output of the fiber from the HET fiber instrument feed (FIF). A pair of achromats optimized for the 1-µm-region images the fiber from the FIF, as well as one from the 49 calibration system, on a 0.1-mm slit. The diverging beam is folded on the way to the collimator which directs the light to a 71.5° blaze angle echelle grating. While undersized, it is twice as efficient as that previously used in Ramsey et al. (2008). This R3 echelle operates in-plane at an off-Littrow angle of ~5° to achieve high dispersion. A 150-l/mm, 5.46° blaze cross disperser optimized for the Y-band between 0.95 and 1.10 µm provides the spectral format in Figure 2-2. We use the same camera system as described in detail in Ramsey et al. (2008), which uses a parabolic mirror with a weak coma correcting lens directly before the NIR dewar. The camera beam is folded, which allows the dewar to be mounted vertically and the detector coolant (liquid nitrogen) to constantly cover the cold plate on which the detector is mounted. Figure 2-3 displays Pathfinder on the optical bench while at HET during our engineering test run. Figure 2-1: Schematic layout of the Pathfinder spectrograph (Ramsey et al. 2010). 50 Figure 2-2: Simultaneous exposure of science and calibration fibers (Ramsey et al. 2010). 51 Figure 2-3: Pathfinder on the optical bench during an engineering test run at HET. Light from the HET and the calibration system is fiber-fed into a slit mechanism (1), bounces off a gold fold mirror (2), reflects off the collimating mirror (3), is dispersed off the grating (covered, 4), is cross-dispersed off the cross disperser (5), reflects off the camera mirror (6), and reflects off a gold mirror (not visible, 7) into the dewar, where approximately 0.2% of the light from the telescope lands on the detector (8) (Redman 2011). 2.2 NIR Detector Pathfinder's detector is a Hawaii 1 1024 × 1024 NIR array with 0.018-mm pixels. As illustrated in Figure 2-4, this array is undersized to cover a full echelle free spectral range of only ~33% of an order. HPF will use a Teledyne H2RG focal plane array and provide nearly full Y-band coverage, but will also require a new refractive camera. Figure 2-4 depicts the free spectral range of the orders in the Y- and J-bands for our current detector, which is the same size as an H1R, and a H2RG, which will be used in HPF. 52 Since Pathfinder is a warm-pupil spectrograph, thermal blocking filters are needed in the dewar to keep the detector from being swamped by the large thermal radiation background at the red end of its sensitivity (~2.5 µm). This is accomplished with a custom Y-band filter, an additional commercial grade thermal blocking filter, and a PK50 glass filter. With this setup, the un-cooled instrument sees a background of ~3 e/s/pixel, easily enabling 10-minute exposures with the HET. Figure 2-4: Free spectral range of the orders in the Y- and J-bands for Pathfinder's crossdispersed echelle configuration, along with the coverage for our current detector (same size as an H1R) and an H2RG, which will be used in HPF (Ramsey et al. 2010). 2.3 Calibration System The major change in the calibration system from that described in Ramsey et al (2008) is that a U-Ne hollow-cathode lamp is employed instead of a Th-Ar lamp. Since 53 Th-Ar lamps have weak thorium lines and strong argon lines (with are not stable at 1 m/s) in the NIR, we have used Pathfinder to explore uranium hollow-cathode lamps. These tests have proven that lamps with neon as a filler gas have fewer bright lines in the NIR Y-band, where we are currently focusing, than argon. Fewer bright lines reduces the negative effects of scattered light and pixel saturation. Our experiments with uranium (U) lamps convincingly show that U has a larger number of emission lines in the Y-band, making it an attractive alternative to thorium (see Figure 2-5). Uranium (238U) fulfills many of the requirements needed of an element in a hollow-cathode lamp for wavelength calibration: it is a heavy element (heavier than thorium), has zero nuclear spin, and has a long half-life. It does have the disadvantage of a very small amount (~0.7%) of 235U in all naturally occurring uranium, which may lead to very subtle isotope shifts between different lamps. A U-Ne hollow-cathode lamp is now the primary calibration lamp for Pathfinder to enable accurate wavelength calibration as well as simultaneous monitoring of the instrument drift. We hope to extend our study of U-Ne to the J- and H-band to verify that U has more lines than Th in these bands, as we expect (Redman 2011; Redman et al. 2012; Mahadevan et al. 2011). A calibration fiber, fed with light from the U-Ne hollow-cathode lamp, is simultaneously fed to the spectrograph in parallel with the star fiber from the FIF. This simultaneous precision emission line reference enables monitoring of the instrument drift. A commercial paint mixer attached to the fibers enables the mitigation of modal noise effects, which are significant in the NIR in our system. 54 Figure 2-5: Comparison of U and Th line densities in the Y-band (Ramsey et al. 2010). 2.4 Preliminary Results from the Hobby-Eberly Telescope While Pathfinder has achieved ~7 m/s using solar light in tests conducted at Davey Lab (Ramsey et al. 2008), that was over hours and with a relatively large amplitude signal of ~300 m/sec. With the goal of 1–3 m/s over a time scale of months, it is critical to conduct telescope observation on real stellar targets. Therefore, Pathfinder was installed at HET in March 2010 in preparation for two test runs. For these initial runs, two relatively bright, stable RV targets were chosen. The first was the giant star HD106714, which is known to be stable to ~10 m/s. As shown in Figure 2-6, binned velocities from 3 nights of observing this star show a scatter of 11.3 m/s. Including the first two nights (where the instrument was still thermally stabilizing), the star exhibited an RV scatter of ~23 m/s for all seven nights during the May run. The second target, Sigma Draconis, is also a RV stable star. The binned velocities for Sigma Draconis show a scatter of ~16.8 m/s, including data from all nights and all observations in May. All 55 observations were taken in the same region in the Y-band. Even in this testbed form, we demonstrated the ability to achieve 10–20 m/s on stars. Figure 2-6: Velocity variation for HD106714 over the May 2010 test run (Ramsey et al. 2010). Subsequent Pathfinder tests at HET in August 2010 were performed using a laser frequency comb developed at the National Institute of Standards and Technology (NIST) labs in Boulder, Colorado. In lab measurements, this LFC demonstrated an absolute accuracy of better than 2 × 10-10 from 1450–1630 nm, corresponding to an RV precision of 6 cm/s (Ycas et al. 2012). As illustrated in Figure 2-7, three 300-µm core multimode fibers feed Pathfinder. One fiber couples light from the telescope, while the other two are used for calibration. Figure 2-8 contains a schematic of the August 2010 test configuration, with beam splitters to direct the different calibration sources into the two calibration fibers. For a detailed description of the LFC test configuration, please see Ycas et al. (2012). During the two weeks of testing, 648 spectra with 5-minute integration times were taken of 91 astronomical objects using Pathfinder's H-band configuration. Figure 2-9 56 displays a spectral image for HD168723, along with its extracted spectra. The LFC provided a uniform series of high SNR emission lines, resulting in an H-band RV precision of ~10 m/s, encouraging results for an uncooled instrument testbed. These LFC tests are the basis for developing an LFC for the HPF spectrograph. However, modal noise posed a limitation for these LFC tests. After the LFC test run, Pathfinder was returned to Penn State for a planned upgrade to Pathfinder II by incorporating what we have learned. We will substantially increase the efficiency and stability and a planned replacement of the current detector to a H2RG will significantly increase the wavelength coverage as well. Our goal with Pathfinder II is to attempt to achieve <5 m/s precision in the NIR, which will be the final step before the implementation of a facility-class instrument. 57 Figure 2-7: Top: The three 300-μm fibers, separated by 125-μm (outer diameter) fibers that acted as spacers for the larger fibers. This fiber pseudo-slit was assembled and polished on-site. From left to right, these are the secondary calibration, primary calibration, and science (stellar) fibers. Bottom: The same fibers, re-imaged onto a 100μm slit. In this image, the primary calibration fiber is transmitting light. The slit was aligned to optimize the orientation of the primary calibration and HET fibers (Redman 2011). 58 Figure 2-8: Schematic of coupling calibration sources into Pathfinder (Ycas et al. 2012). Figure 2-9: Spectra of the star HD16823 along with LFC calibration. Echelle orders 37– 39 are visible, with wavelengths of 1536.6–1543.9 nm, 1577.1–1584.6 nm, and 1619.7– 1627.4 nm, respectively (Ycas et al. 2012). 59 Chapter 3 Preliminary Laser Diode Modal Noise Tests The following tests represent the preliminary step in characterizing and mitigating modal noise in Penn State's Pathfinder spectrograph. These early tests highlight the experimental subtleties of measuring modal noise in the visible and NIR. Necessary improvements were made in our modal noise test configuration and data analysis techniques based on these results. 3.1 Test Procedure For these preliminary experiments, modal noise was measured at 635-nm and 1550-nm using Thor Labs laser diodes, which are pigtailed to 1 meter long single mode fibers with an FC/PC connector. Table 3-1 provides the laser diode specifications. For all tests, a Thor Labs ITC 4001 Laser Diode/Temperature Controller was used to set the laser diode currents and stabilize the diode temperatures at 23 °C. The 635 nm and 1550 nm laser diodes were alternately coupled to an Edmund Optics 1 meter long 0.22 NA VIS/NIR 200 µm core multimode fiber (NT57-748) with an FC/PC connector. Using this setup, the input to the multimode fiber is only the single fundamental mode of coherent laser light at the given laser diode's wavelength (in reality, those wavelengths within the laser diode's finite bandwidth). Any modal noise detected in the multimode fiber's far- 60 field pattern will be due to the multimode fiber's variations and irregularities, comparable to Pathfinder's actual test configuration with its multimode science fiber. Table 3-1: Thor Labs laser diodes used in preliminary modal noise testing. LPS-635-FC LPS-1550-FC Wavelength 635 nm 1550 nm Spectral Bandwidth Δλ ~ 0.08 nm Δλ ~ 1.5 nm Output Power 2.5 mW 1.5 mW The multimode fiber was mounted and the output light collimated so that the beam could be detected on a Xenics Xeva-1.7-320 NIR camera, which has a 256 × 320 pixel array with peak sensitivity from 900–1700 nm (22% QE at 635 nm and 84% QE at 1550 nm, see Figure 3-1). A Newport motorized iris (Model 62280) reduced the collimated beam diameter so that the beam spot would properly fit on the NIR camera's detector. An FMC Syntron PowerPulse electromagnetic paint mixer was used to vibrate the multimode fiber at 60 Hz, so that static and vibrated results could be compared for 635 nm and 1550 nm. Figure 3-2 shows the vibrator attached to the multimode fiber. Figure 3-1: Xenics NIR camera quantum efficiency. 61 Figure 3-2: Vibrator attached to 200-µm multimode fiber. Flat field frames were taken by placing a Dolan-Jenner Fiber-Lite area backlight in front of the 100% opened (36-mm-diameter) motorized iris to allow uniform illumination across the detector. For the laser diode tests, the motorized iris was set to 5% open (3.7 mm diameter) so that the beam spot would properly fit on the detector. In addition, images were multiplied by a circular mask to eliminate the effects of scattered background light lying outside of the main laser beam. Frame integration times were adjusted and neutral density filters were used to limit the maximum counts to roughly half of the NIR detector's dynamic range. An integration time of 0.6 ms was used for the 1550 nm-tests, while 300 ms was used for the 635-nm tests. Three individual sets of 10 frames in sequence were taken for both the 635-nm and 1550-nm laser diodes. The first sets consisted of directly imaging the laser diode's single mode fiber, while the second sets involved imaging the output of the multimode fiber under static conditions while connected to the laser diode's single mode fiber. For the third sets, the laser diode's single mode fiber was connected to the multimode fiber, 62 which was vibrated at 60 Hz. All 10 frames from each individual set were then medianaveraged to produce one combined frame for each test condition. 3.2 Results and Analysis Figure 3-3 displays the results of the 635-nm single mode fiber test. As expected, the 635-nm laser diode produced a Gaussian profile of the single mode in the fiber, i.e., the fundamental mode. This serves as a comparison to the modal variations present in the static and vibrated multimode fiber results. Figure 3-3: 635-nm single mode fiber test: (left) 635-nm single mode fiber (average of 10 frames), (right) 3D intensity plot. Figure 3-4 features the results of the 635-nm laser diode static and vibrated tests. For the static case, large modal patterns clearly can be seen across the fiber's far-field pattern. However, vibration of the multimode fiber has largely removed the modal patterns in the vibrated frame. Not only is there less modal variation present, but the 63 remaining variation also has smaller amplitude fluctuations. It is the reduction of the amplitude and time variation of these modal patterns that will improve the spectrograph illumination for Pathfinder. Figure 3-4: 635-nm static and vibrated fiber tests: (left) 635-nm static image (average of 10 frames), (right) 635-nm vibrated image (average of 10 frames) Figure 3-5 shows the 1550-nm laser diode static and vibrated fiber test results. A greater number of speckles are readily visible in the fiber far-field pattern and with a greater range of intensities than the 635-nm case. Once again, vibrating the multimode fiber has the effect of smoothing over the modal intensities, though variations remain. 64 Figure 3-5: 1550-nm static and vibrated fiber tests: (left) 1550-nm static test (average of 10 frames), (right) 1550-nm vibrated test (average of 10 frames). Comparing Figures 3-4 and 3-5, it is clear that the speckle pattern in the 1550 nm case is more prominent. As Equation 1.6 details, the 635-nm laser diode should excite roughly 6 times more modes than the 1550-nm laser diode. Since many more modes are excited in the 635-nm test, the MPD presents itself as a relatively uniform beam spot with low spatial frequency changes and a higher SNR. However, fewer modes are excited in the 1550-nm test, and so the interference of the individual modes is readily visible as a speckle pattern in the fiber's far-field output. The amplitude and time variation of this NIR speckle pattern will have worse implications for stable spectrograph illumination than the visible case in Figure 3-4. These results highlight the particular challenge of detecting exoplanets around M dwarfs and the importance of our research in reducing modal noise in our NIR observations. The visual representations of these test results are helpful, but a spatial frequency analysis is required to fully characterize modal variations. Figure 3-6 displays the 2D rotationally-averaged spatial frequency power spectrum plots for all 4 test cases, with 65 normalized power and spatial frequency axes. The Xenics NIR camera has 0.03-mm pixels and so spatial frequencies are given in cycles/mm, or simply mm–1, and plotted out to the Nyquist frequency (1 cycle / 2 pixels = 16.67 mm–1). Since fibers are vibrated to scramble the power in the various excited modes and homogenize the far-field speckle pattern, a perfectly scrambled multimode fiber illumination pattern would only consist of the DC frequency (0 mm–1), i.e., the far-field pattern would appear as a beam spot of constant intensity and the power spectra would have a delta function spike at 0 mm–1. In Figure 3-6, note that the static and vibrated results do contain the most power at 0 mm–1, but there is significant power in the higher spatial frequencies as well. The 1550-nm static and vibrated test cases have approximately 2–3 orders of magnitude more power than the corresponding spatial frequencies of the 635-nm test cases. This is indicative of the large speckle contrast and modal variations present in the 1550-nm images of Figure 3-5 and, conversely, the relative homogeneity and lower spatial frequency modal changes in the 635-nm images of Figure 3-4. In particular, our modal noise tests aim to reduce the lower spatial frequencies, as they are the most problematic for stable spectrograph illumination. From Figure 3-6, a 50% reduction in low spatial frequencies (0.5–1.5 mm–1) can be discerned going from the 635-nm static to the vibrated case, whereas the power remains relatively constant at all other spatial frequencies. This verifies the visual modal reduction witnessed in Figure 3-4. The reduction of low spatial frequencies is more evident in the 1550-nm power spectra over a wider spatial frequency expanse. The power in the 0.33–12.5 mm–1 spatial frequency range is reduced by 60% from the 1550 66 nm static to vibrated case. Clearly, vibration has a more dramatic impact for the 1550nm laser diode, which is encouraging for our future NIR modal noise tests. These tests proved that modal variations were reduced by vibration in both cases, but longer integration times still needed to be tested to fully characterize modal noise reduction for the visible and NIR. The integration times for these individual static and vibrated frames ranged from 0.6–300 ms, but integration times on the order of 1 second are needed to more closely match Pathfinder's much longer exposure time. Longer integration times allow for more intensity averaging over a much greater number of vibration cycles. 67 Figure 3-6: 2D spatial frequency power spectra: 635-nm static (solid blue line), 635-nm vibrated (solid red line); 1550-nm static (dashed blue line), 1550-nm vibrated (dashed red line). 3.3 Optical System and Image Analysis Improvements A number of improvements were identified in our test configuration based on the results of these preliminary tests. The following sub-sections describe these changes, which were necessary to implement before conducting the more comprehensive modal noise testing described in Chapter 4. 68 3.3.1 Camera Calibration A crucial aspect of any optical system is proper calibration, since poor calibration will ultimately produce low-integrity images that lead to erroneous data analysis. Xeneth, the software program used to operate and control the Xenics NIR camera, allows users to recalibrate the camera by creating calibration packs that can be loaded depending on the user's desired settings (e.g., setting different integration times). A necessary component of creating new calibration packs is producing uniform illumination for flat field frames, since "dark" and "grey" frames (i.e., flat field frames with illumination just above the dark threshold and flat field frames at roughly ¾ of the detector's dynamic range, respectively) are needed for Xeneth's recalibration process. In addition, proper flat field frames are needed for correcting pixel response variations in actual test frames. Upon further testing, it was found that the Dolan-Jenner area backlight (see Figure 3-7) produced relatively poor flat field frames with large-scale spatial structure. Modifying the setup by placing a diffuser in a filter holder directly in front of the camera dramatically improved the flat field results, since the diffuser removed any inhomogeneities in the incident light pattern. Figure 3-8 shows an example of an improved flat field frame, while Figure 3-9 displays a normalized master flat field frame created by median averaging 101 individual flat field frames. Using this improved recalibration technique, new calibration packs were developed for longer integration times (up to 3 sec), which allowed for tests more closely resembling Pathfinder's ~3 min integration time. 69 Figure 3-7: Dolan-Jenner area backlight used for flat fielding. Figure 3-8: Improved flat field frame. 70 Figure 3-9: Normalized master flat field frame. 3.3.2 Dust Mitigation Dust accumulation on the glass window covering the Xenics NIR camera's detector chamber or on the neutral density filters needed to attenuate the light from the laser diodes resulted in image degradation (see Figure 3-10). The small size of the dust particles introduced unwanted high spatial frequencies into 2D FFT and power spectra plots. Since the flat field frames were taken with diffuse light and the test frames were taken with collimated light from the laser diodes, the flat fields could not be used to correct for the dust particles. Therefore, we had to mitigate the dust accumulation by cleaning the glass window with compressed air and a cotton applicator the best that we could. A 560-nm red filter was taped over the camera's aperture above the glass window 71 to minimize any further dust accumulation. The red filter could then be removed when conducting tests. Figure 3-10: Sample image highlighting dust accumulation. 3.3.3 Diffraction Effects Upon further testing, it was noticed that setting the motorized iris 5% open (3.7 mm diameter) introduced diffraction effects in the fiber's far-field pattern imaged on the detector, as shown in Figure 3-11. Since these wavelength-dependent diffraction effects introduced unwanted spatial frequencies, which impact modal noise measurements, it was decided to keep the motorized iris 100% open and to place a converging lens in front of the camera. The size of the far-field pattern on the detector could then be controlled by moving the camera closer or farther away from the lens on a linear actuator. These 72 changes in the test configuration eliminated the diffraction effects, while keeping the relative scaling of the modal variations in the far-field pattern. Figure 3-11: Diffraction effects caused by motorized iris. 3.3.4 Optical Coupling Further testing also revealed the intensity variations in the fiber far-field pattern resulting from the uneven illumination of the multimode fiber by the single fiber at the FC/PC connector. To remove this mechanical dependency, the test configuration was altered so that the laser diode single mode fibers could be optically coupled to the multimode test fiber. Figures 3-12 and 3-13 display the optical coupling setup. A fiber mount for the laser diode single mode fibers was placed at the focus of a converging lens. Another fiber mount for the multimode fiber was placed on a linear actuator at the 73 opposite focus of the converging lens. This allowed the multimode fiber mount to be moved closer or farther from the lens in order to properly illuminate the input end of the fiber. The neutral density filters were also moved from the collimated beam path to the optical coupling location. Removing the filters from the collimated beam and placing them before the multimode fiber helped reduce the fringing effects caused by the filters. Single mode fiber Lens Multimode fiber Figure 3-12: Single mode fiber optically coupled to the multimode fiber. 74 Figure 3-13: Schematic of optical coupling setup within the overall test configuration. 3.4 Conclusions The preliminary laser diode tests proved that modal variations were reduced by vibrating the multimode fiber. The 635-nm laser diode static fiber image (see Figure 3-4) contained large-scale modal structure with minor intensity variations. This was consistent with Equation 1.6, which states that shorter wavelength light will excite a larger number of modes in an optical fiber. The high number of excited modes interfered at the fiber output, producing a largely homogeneous image. For the 1550-nm case, the static fiber image (see Figure 3-5) contained much larger intensity variations due to the lower number of excited modes present in the fiber. The 1550-nm tests also produced higher power across the entire range of spatial frequencies than the 635-nm cases, consistent with the larger amount of modal intensity variations clearly visible in the 1550 75 -nm far-field speckle pattern. Though vibration reduced the modal variations for both the 635-nm and 1550-nm cases, this reduction was more pronounced at 1550 nm than 635 nm, reducing power across a significant range of spatial frequencies. 76 Chapter 4 Visible and Near-Infrared Modal Noise Tests With all of the necessary test configuration improvements in place, more comprehensive modal noise tests were ready to be conducted. Chapter 4 contains a description of the test procedures and analysis of the results. 4.1 Laser Diode and Narrow-Band Laser Tests The goal of these laser diode and narrow-band laser tests is to provide a more comprehensive examination of modal noise in the visible and NIR. More laser diodes were tested at different wavelengths, while narrow-band lasers provided a more realistic comparison to Pathfinder's resolving power. 4.1.1 Test Procedure Modal noise was tested using 635-nm, 830-nm, 1310-nm, and 1550-nm Thor Labs laser diodes, a 632.8-nm Metrologic ML 810 HeNe laser (1.0 mW), and a 1550-nm Acronym Fiber Optics, Inc. (Item # DFB-C10) narrow-band, distributed feedback (DFB) laser. Table 4-1 contains the laser and laser diode specifications. The Thors Labs ITC 4001 Laser Diode/Temperature Controller was once again used to set the laser diode currents and stabilize the diode temperatures at 23 °C. The lasers and laser diodes were 77 optically coupled to a 0.22 NA, 30-meter-long Polymicro multimode fiber with a 200-µm diameter silica core, 20-µm thick cladding, and 40-µm thick polyimide buffer. This longer fiber is a better analog to Pathfinder's ~30-meter science fiber than the shorter multimode fiber used in the preliminary tests. Table 4-1: Laser and laser diode specification (LD = laser diode). Wavelength Spectral Bandwidth Resolving Power (λ/Δλ) Output Power LD 635 nm 0.08 nm 7,938 2.5 mW Laser 632.8 nm 0.01 nm 63,280 0.8 mW LD 830 nm <0.1 nm 8,300 10.0 mW LD 1310 nm ~1 nm 1,310 2.5 mW LD 1550 nm 1.5 nm 1,033 1.5 mW Laser 1550 nm 0.08 pm 19,375,000 20 mW 1550-nm narrow-band laser 632.8-nm HeNe laser Laser diode mounts Figure 4-1: Test setup for comprehensive laser and laser diode tests. 78 In addition to the Xenics camera used for the NIR tests, a 1056 × 1027 pixel Finger Lakes Instrumentation (FLI) PL4710 deep depletion CCD camera with 300–1000 nm wavelength coverage was used to provide higher-resolution results in the visible and NIR. The FLI's E2V CCD has 13-µm pixels, compared to the 30-µm pixels of the Xenics camera's NIR array. Figure 4-2 displays the quantum efficiency of the FLI camera across the visible and NIR spectrum. Figure 4-2: FLI PL4710 deep depletion camera quantum efficiency (black line). Three individual sets of 10 frames in sequence were taken for each of the four laser diodes and two lasers. The first sets were taken with the multimode fiber static (referred to as the static test), the second sets with the multimode fiber vibrated at 60 Hz by the electromagnetic paint mixer (mechanical vibration test), and the third sets with the multimode fiber agitated by hand at ~1–2 Hz with an amplitude of ~10–15 cm (hand agitation test). For the hand agitation tests, the multimode fiber was held with two hands approximately 25 cm apart. Each hand was alternatively moved up and down 180° out of 79 phase at the specified frequency and the given amplitude. Individual test frames were taken with a 3-second integration time to average over a significant number of vibrational periods and to provide a sufficient SNR. For the 635-nm and 830-nm laser diodes, individual sets were taken with both the Xenics and FLI cameras to provide a performance comparison between the two cameras in the visible and NIR. Background frames were median-averaged and subtracted from all of the FLI frames to remove the effects of ambient and scattered light in the visible. As before, each individual set of 10 frames was median-averaged to provide one combined frame for each test condition. 4.1.2 Results and Analysis Figure 4-3 displays images and rotationally-averaged, log-scaled 2D FFT plots for the 635-nm laser diode in both the Xenics and FLI camera static test cases. The FLI camera offers a much better representation of the modes present in the multimode fiber. The smaller 13-µm pixels of the FLI camera provide over 5 times the spatial resolution of the Xenics camera's 30-µm pixels, which blur much of the modal pattern (this is also witnessed in the 830-nm Xenics test case). The 2D FFT plots clearly show that the FLI static image has much more power in the 0–15 mm–1 spatial frequency range compared to the Xenics static image, despite both cameras imaging the same 635-nm laser diode multimode fiber output (note that the FLI camera's Nyquist frequency is 1 cycle / 2 pixels = 38.46 mm–1). The total amount of modal blurring due to the Xenics camera's undersampling of the 635-nm modal pattern was not appreciated in the preliminary tests, since there were no higher-resolution visible camera results for comparison. Since the 80 comprehensive modal noise analysis in this chapter is highly dependent on the ability to truly capture modal changes, my analysis only focuses on the higher-resolution FLI results for the 635-nm and 830-nm laser diodes. Figure 4-3: 635-nm static fiber tests: (top left) Xenics camera image, (top right) Xenics camera image 2D FFT, (bottom left) FLI camera image (bottom right) FLI camera image 2D FFT. Despite the fact that some of this effect is undoubtedly due to undersampling, there appears to be a more considerable factor at work. The Xenics camera may have an inherently lower modulation transfer function (MTF), particularly at higher spatial 81 frequencies. A lower MTF for the Xenics camera would also lead to blurring, even without undersampling. The Xenics and FLI camera images in Figure 4-3 reveal low spatial frequency inhomogeneities. These inhomogeneities arise due to the imperfect optical coupling and are dependent on the specific input coupling geometry, which determines the amount of power in each excited mode. However, optical coupling is an improvement over mechanical coupling with FC/PC connectors because of the ability to adjust the fiber input illumination. In any case, the input illumination of optical fibers in astronomical spectrographs is never perfect, so in actuality, these coupling inhomogeneities provide more realistic test conditions. Figure 4-4 displays the mechanical and hand agitation images and log-scaled 2D FFT plots for the 635-nm laser diode with the FLI camera, while Figure 4-5 contains their power spectra plots. All power spectra plots presented in this chapter are normalized to their individual DC values to remove dependencies on the intrinsic dynamic range difference between the Xenics and FLI cameras and the different combinations of neutral density filters needed to keep maximum intensities at roughly one-half to three-quarters of these dynamic ranges. It is readily apparent in the 635-nm laser diode test that the mechanical vibration significantly reduced the power in the 1–20 mm–1 spatial frequencies (by up to 96%), but that the hand agitation has an even greater effect (up to 99.5%), particularly in the 1–15 mm–1 range. Therefore, the amplitude of fiber vibration appears to play a substantial role in reducing the power of spatial frequencies in modal patterns. While the mechanical vibration amplitude was only a few millimeters, the hand agitation amplitude of ~10 cm provided enough stress to redistribute the modal power 82 weights, reducing the lower spatial frequency power by almost an order of magnitude more. The amplitude dependence of modal pattern reduction appears to outweigh the dependence on vibration frequency, since the hand agitation was at a much lower frequency than the mechanical vibration. Despite the enhanced scrambling gained through mechanical and hand agitation, it is interesting to note that the static, mechanical vibration, and hand agitation images all have practically the same amount of power in spatial frequencies > 20 mm–1. This seems to indicate that the white noise limit has been reached at these spatial frequencies. The large-scale inhomogeneities due to the optical coupling geometry become slightly more prominent proceeding from the static test to the hand agitation test and minor fringing effects become visible as well. This seems to signify that neither small nor large-scale vibration is able to homogenize the entire image and it is these large-scale inhomogeneities that will be dominant at the telescope. 83 Figure 4-4: 635-nm laser diode vibration tests (FLI camera): (top left) mechanical vibration image, (top right) mechanical vibration image 2D FFT, (bottom left) hand agitation image, (bottom right) hand agitation image 2D FFT. 84 Figure 4-5: 635-nm laser diode tests power spectra (FLI camera): (blue line) static image power spectrum, (black line) mechanical vibration image power spectrum, (red line) hand agitation image power spectrum. Although laser diodes are useful for modal noise testing, the laser diode spectral resolving powers listed in Table 4-1 are well below Pathfinder's resolving power of ~ 50,000. Consequently, the HeNe and narrow-band laser test results presented in this chapter are important for measuring modal noise at a resolving power comparable to Pathfinder. Figure 4-6 shows the 632.8-nm HeNe laser static, mechanical vibration, and hand agitation test results with the FLI camera. Fewer speckles are visible in the HeNe 85 laser's static image compared to the 635-nm laser diode static image in Figure 4-3, since the laser's smaller bandwidth leads to fewer excited modes. As described in §1.5.3.2, modal noise becomes a significant problem for coherent light. This fact becomes even more apparent when considering the 632.8-nm HeNe laser static power spectrum in Figure 4-7, where the power is roughly an order of magnitude higher across the entire spatial frequency range compared to the 635-nm laser diode static power spectrum. Vibration of the multimode fiber had a rather unique result for the 632.8-nm HeNe laser tests. While mechanical vibration reduced the power in the 0.58–35 mm–1 spatial frequency range by about 50–70%, the hand agitation was two-to-three orders of magnitude more effective, maxing out at a reduction of 99.93% at 10 mm–1. In fact, comparing Figures 4-5 and 4-7, hand agitation reduces the spatial frequency power to almost the exact same levels across the entire spatial frequency range for both the HeNe laser and 635-nm laser diode tests. So, although the smaller bandwidth of the HeNe laser initially produces a higher power modal pattern in the static image, larger amplitude vibration of the multimode fiber effectively nullifies this effect relative to the wider bandwidth 635-nm laser diode. 86 Figure 4-6: 632.8-nm HeNe laser tests (FLI camera): (top left) static image, (top right) static image 2D FFT, (middle left) mechanical vibration image, (middle left) mechanical vibration image 2D FFT, (bottom left) hand agitation image, (bottom right) hand agitation image 2D FFT. 87 Figure 4-7: 632.8-nm HeNe laser tests power spectra (FLI camera): (blue line) static image power spectrum, (black line) mechanical vibration image power spectrum, (red line) hand agitation image power spectrum. Moving closer to the Y-band, which will be the focus of HPF, Figure 4-8 displays the 830-nm laser diode test results with the FLI camera, whereas Figure 4-9 contains the corresponding power spectra plots. Once again, hand agitation has a greater spatial frequency power reduction than mechanical vibration. Mechanical vibration reduces spatial frequency power by 40–70% in the 0.87–25 mm–1 range, while hand agitation diminishes spatial frequency power by 98–99%, with a maximum decrease of 99.7% at 7.596 mm–1. Although mechanical vibration had a greater effect in the 635 nm laser diode test compared to the 830-nm laser diode test, hand agitation reduced spatial 88 frequency powers to comparable, if not slightly lower, levels for the 830-nm laser diode compared to the 635-nm laser diode tests. As with the 635-nm laser diode tests, neither mechanical nor hand agitation has any effect on higher spatial frequencies (in this case, > 25 mm–1). Large-scale inhomogeneities are more readily visible in the 830-nm laser diode vibrated images compared to the static image, verifying that these modal structures are independent of fiber agitation, similar to the 635-nm laser diode tests. 89 Figure 4-8: 830-nm laser diode tests (FLI camera): (top left) static image, (top right) static image 2D FFT, (middle left) mechanical vibration image, (middle left) mechanical vibration image 2D FFT, (bottom left) hand agitation image, (bottom right) hand agitation image 2D FFT. 90 Figure 4-9: 830-nm laser diode tests power spectra (FLI camera): (blue line) static image power spectrum, (black line) mechanical vibration image power spectrum, (red line) hand agitation image power spectrum. Now squarely in the NIR regime, the remaining 1310-nm and 1550-nm tests results utilized the Xenics camera. Although the Xenics camera has a lower spatial resolution than the FLI camera, the results from the previous three tests have proven that the most substantial spatial frequency reduction takes place below 15 mm–1, which is within the spatial resolution of the Xenics camera. Figure 4-10 features the 1310-nm laser diode test results and Figure 4-11 displays the corresponding power spectra plots. Vibration begins to reduce spatial frequency 91 power at about 1 mm–1 and continues all the way to the Nyquist frequency of 16.67 mm–1. Mechanical vibration reduced spatial frequencies by 70–80% over much of this range, though the reduction is noticeably less >11 mm–1. Hand agitation improved the spatial frequency reduction to 90–97%, but large-scale inhomogeneities remain. Note that all three of the 1310-nm power spectra have one to two orders of magnitude more power than the corresponding shorter wavelength power spectra. 92 Figure 4-10: 1310-nm laser diode tests (Xenics camera): (top left) static image, (top right) static image 2D FFT, (middle left) mechanical vibration image, (middle left) mechanical vibration image 2D FFT, (bottom left) hand agitation image, (bottom right) hand agitation image 2D FFT. 93 Figure 4-11: 1310-nm laser diode tests power spectra (Xenics camera) – (blue line) static image power spectrum, (black line) mechanical vibration image power spectrum, (red line) hand agitation image power spectrum. Figures 4-12 and 4-13 show the 1550-nm laser diode test images and power spectra plots, respectively. Fewer speckles are visible in these images compared to the previous tests, consistent with Equation 1.6, since longer wavelengths excited fewer modes. Unlike the 1310-nm laser diode tests, vibration reduces spatial frequency power consistently across the entire spatial frequency, including higher spatial frequencies. Mechanical vibration reduces all spatial frequencies by 70–80%, but hand agitation increases this amount to 93–99.7%. Large inhomogeneities are visible in the lower left of all three images and fringing effects are noticeable in the hand agitation image. 94 Figure 4-12: 1550-nm laser diode tests (Xenics camera): (top left) static image, (top right) static image 2D FFT, (middle left) mechanical vibration image, (middle left) mechanical vibration image 2D FFT, (bottom left) hand agitation image, (bottom right) hand agitation image 2D FFT. 95 Figure 4-13: 1550-nm laser diode tests power spectra (Xenics camera): (blue line) static image power spectrum, (black line) mechanical vibration image power spectrum, (red line) hand agitation image power spectrum. The 1550-nm narrow-band laser provides a higher resolving power comparison to the 1550-nm laser diode results. The sparse speckle pattern of the 1550-nm narrow-band laser static image in Figure 4-14 is a testament to the low number of excited modes. Analyzing the 1550-nm narrow-band laser power spectra plots in Figure 4-15, it is apparent that mechanical vibration once again reduces spatial frequency power by about 70–80% across the 0.67–16.67 mm–1 range, while hand agitation lessens spatial frequency power by 98–99% until >10 mm–1, where the decrease is 90%. The large-scale inhomogeneities due to the coupling geometry play a major role in this test case with the low number of excited modes. The static image has a small region with very few modes 96 and vibration proceeds to expose this region even more. The white noise limit is never reached, so these large scale regions dominate even at higher spatial frequencies. 97 Figure 4-14: 1550-nm narrow-band laser tests (Xenics camera): (top left) static image, (top right) static image 2D FFT, (middle left) mechanical vibration image, (middle left) mechanical vibration image 2D FFT, (bottom left) hand agitation image, (bottom right) hand agitation image 2D FFT. 98 Figure 4-15: 1550-nm narrow-band laser tests power spectra (Xenics camera): (blue line) static image power spectrum, (black line) mechanical vibration image power spectrum, (red line) hand agitation image power spectrum. Comparing Figures 4-13 and 4-15, it is clear that the 1550-nm narrow-band laser and laser diode have roughly comparable levels of power at all spatial frequencies. Hand agitation reduces these levels, but the reduced power is still 3–4 times higher across the entire spatial frequency range in the 1550-nm narrow-band laser test relative to the 1550nm laser diode test. This is a slight deviation from the 632.8-nm HeNe laser and 635-nm laser tests, where hand agitation reduced the spatial frequency power in both static images to almost exactly the same levels throughout the total spatial frequency range. The 1550-nm laser diode and narrow-band laser tests reveal another noteworthy aspect of modal noise reduction in the visible and NIR. As depicted in Figures 4-13 and 99 4-15, the 1550-nm laser diode and narrow-band laser static image power spectra contain one to two orders of magnitude more power across the entire spatial frequency range than the corresponding static plots of the previous shorter wavelength tests, particularly the tests at 635 nm and 830 nm. Despite the power reduction due to vibration, the 1550-nm laser diode and narrow-band laser vibrated and agitated power spectra still stay roughly one-to-two magnitudes greater than their vibrated and agitated power spectra counterparts in the previous tests. In fact, both of the 1550-nm hand-agitated power spectra contain more power than even the static power spectra at all of the shorter wavelengths. This may be largely due to the much larger contrast differences in the 1550-nm vibrated images caused by the large-scale inhomogeneities. In any case, larger power levels at higher spatial frequencies lead to more modal noise when the multimode fiber experiences internal or external stresses in an astronomical spectrograph. This result emphasizes the importance of reducing modal noise as much as possible in the NIR for M dwarf exoplanet detection. Consequently, larger amplitude vibration will be essential in reducing modal noise in HPF. In addition to the images, 2D FFTs, and power spectra previously examined, speckle contrast is another useful measurement for characterizing speckle reduction. According to speckle theory, the speckle contrast of an intensity pattern is defined by: C I , (4.1) where C is the speckle contrast, σ is the standard deviation of the intensity pattern, and I is the average intensity (Goodman 1976). Vibration scrambles the amount of power within excited fiber modes, decreasing speckle contrast. A perfectly scrambled far-field 100 speckle pattern, i.e., an image of constant intensity, will have a speckle contrast of zero. Table 4-2 lists the speckle contrasts for the two laser tests and all four laser diode tests, while Figure 4-16 presents these values in graphical form. 101 Table 4-2: Speckle contrasts for laser and laser diode tests. Test Source LD 635 nm Laser 632.8 nm LD 830 nm LD 1310 nm LD 1550 nm Laser 1550 nm Camera FLI FLI FLI Xenics Xenics Xenics Test Type Speckle Contrast Static 0.743 Mechanical vibration 0.669 Hand agitation 0.809 Static 0.848 Mechanical vibration 0.763 Hand agitation 0.756 Static 0.774 Mechanical vibration 0.739 Hand agitation 0.688 Static 0.564 Mechanical vibration 0.578 Hand agitation 0.554 Static 0.930 Mechanical vibration 0.879 Hand agitation 0.672 Static 1.180 Mechanical vibration 1.114 Hand agitation 0.984 102 Figure 4-16: Laser and laser diode test speckle contrasts: (blue) static images, (green) mechanical vibration images, (red) hand agitation images. "LD" indicates laser diode, "F" and "X" denote FLI and Xenics cameras, respectively, and "Laser633" specifies the HeNe laser. From Table 4-1 and Figure 4-16, it is evident that mechanical vibration and hand agitation successively reduce the speckle contrast in almost all test cases. Speckle contrast values tend to increase towards longer wavelengths, the notable exception being the 1310-nm laser diode, which was imaged with the lower resolution Xenics camera. Both the 632.8-nm HeNe laser test and the 1550-nm narrow-band laser tests have higher 103 speckle contrast than their corresponding laser diode tests, once again indicating the more prominent speckle patterns that result from smaller bandwidths, i.e., light that is more highly coherent. However, speckle contrast can be a somewhat misleading measurement, since vibration does not remove all of the large-scale inhomogeneities. These large-scale patterns can have subtle effects on speckle contrast, as demonstrated by the 635-nm laser diode tests, where the speckle contrast increases with hand agitation. Speckle contrast is a useful first-order statistic, but the power spectra tell the true story of modal reduction, with their description of the power at all possible spatial frequencies. 4.1.3 Conclusions The most significant result from these modal noise tests is that, although the electromagnetic paint mixer has been vibrating the multimode science fiber in Pathfinder and reducing modal noise, larger amplitude agitation has a tremendous impact on further reducing modal variations. In all of the test cases, the larger amplitude hand agitation provided a modal reduction of about one-to-two orders of magnitude more than the mechanical vibration, despite being at a much lower vibration frequency. In most tests cases, the mechanical vibration achieved a modal reduction of 50–70%, whereas hand agitation spatial frequency reduction was typically 98–99%. The higher-resolution FLI camera played a useful role in proving that modal reduction can extend much further into the higher spatial frequencies, especially when imaging a source with a high resolving power (e.g., the 632.8-nm HeNe laser). 104 It was shown that the longer wavelength 1310-nm and 1550-nm tests contained more power at all spatial frequencies than the lower wavelength 632.8-nm, 635-nm, and 830-nm tests. In addition, the smaller bandwidth of the lasers produced more spatial frequency power in the multimode fiber's far-field speckle pattern than the wider-band laser diodes. Moving from the visible to the NIR for precision RV measurements and utilizing a spectrograph with a high resolving power, both of these facts contribute unwanted noise to our efforts to measure the centroids of stellar absorption lines. However, although the narrower bandwidth of the 632.8-nm HeNe laser produced more spatial frequency power in the static image compared to the 635-nm laser diode static image, hand agitation nullified this initial disparity, reducing both hand-agitation images to matching spatial frequency powers. Hand agitation was not quite strong enough to produce the same nullifying effect at 1550 nm, but other narrow-band sources will need to be tested to probe this outcome at other NIR wavelengths. Taking the hand agitation results into consideration, we are currently designing a large-amplitude mechanical fiber agitator that will be able to achieve the same vibration frequency and perform the same range of motion as demonstrated here in our hand agitation tests (see Figure 4-17). This mechanical agitator will significantly lessen the impact of modal noise in Pathfinder and in the upcoming HPF. 105 Figure 4-17: Preliminary design of the large-amplitude mechanical agitator (Courtesy of Daniel Anderson and Kyle McCausland). 106 Chapter 5 Conclusions and Future Work After thousands of years of speculation by philosophers and astronomers, we currently stand on the verge of detecting habitable worlds beyond Earth. However, the detection of the first habitable exoplanet will not be the end of this philosophical and scientific journey; it will be the beginning of a new age of discovery. Our team at Penn State has taken major strides in this effort with our Pathfinder prototype spectrograph. The lessons we are learning will be applied to the future facilityclass HPF, but work remains in reducing sources of RV error, particularly modal noise. The research presented in this thesis lays down the groundwork for how we will continue to measure and diminish the effects of modal noise and apply the results to Pathfinder and HPF. The preliminary and comprehensive modal noise tests presented here prove that using laser diodes and narrow-band lasers in a lab bench test setup is an effective way to probe the effects of modal noise. This test setup is much simpler and cheaper than configuring an entire spectrograph for modal noise testing, while still providing meaningful results. Different wavelengths and resolving powers can be easily tested by merely changing the input source to the multimode fiber and various vibration techniques can be explored interchangeably. It was shown that larger amplitude agitation has a very distinct impact on modal variation reduction; in fact, large-amplitude agitation proved to be a few orders of 107 magnitude more effective than small amplitude vibration at every wavelength tested. As a result, our new large-amplitude mechanical agitator will soon be incorporated into our modal noise test setup and will be used with Pathfinder, providing a repeatable solution for the experimental hand agitation tested in this thesis. The large-amplitude mechanical agitator will be able to investigate the effects of higher frequency large-amplitude agitation, since it remains unknown whether increasing the large-amplitude agitation above ~1–2 Hz will have a significant impact. The impact of large-amplitude agitation on fiber lifetime will also be examined to ensure long-term performance. The large-scale intensity inhomogeneities that became more apparent with greater levels of spatial frequency reduction and at longer wavelengths demonstrated that the coupling geometry can considerably alter the multimode fiber's far-field speckle pattern that is ultimately fed into Pathfinder. For precision RV measurements in the NIR, we will have to pay particular attention to stabilizing Pathfinder's multimode input fiber. Upcoming tests with Pathfinder aim to determine the crucial relationship between modal noise reduction and the corresponding shift in spectral line centroids. The Thor Labs laser diodes will be used to illuminate Pathfinder's multimode fiber so that we can measure modal noise at different regions in the visible and NIR. These future tests will provide some of the first quantifiable results linking modal noise reduction in the NIR to improved RV precision. Pathfinder will also enable us to measure the full width at half maximum (FWHM) of the laser diodes, which will allow explicit comparisons to the narrow-band lasers in terms of the number of excited modes and provide an opportunity to test modal theory in a real system. 108 In addition, new experimental studies are beginning to indicate that square and octagonal fibers show promising improved scrambling performance over conventional circular fibers (Chazelas et al. 2010). Though circular fibers will continue to be used on Pathfinder and HPF, we will explore the properties of square and octagonal fibers and determine the benefit of potentially using these fibers with our instruments. 109 References Avila, G. and P. Singh. 2008. "Optical fiber scrambling and light pipes for high accuracy radial velocities measurements." Advanced Optical and Mechanical Technologies in Telescopes and Instrumentation. Edited by Atad-Ettedgui, E. and D. Lemke. Proceedings of the SPIE, Vol. 7018, pp. 70184W-1–70184W-7. Avila, G., Singh, P., and Albertsen, M. 2006. "Photometrical scrambling gain and focal ratio degradation in fibers for astronomical instruments." Ground-based and Airborne Instrumentation for Astronomy. Edited by McLean, I.S. and M. Iye. Proceedings of the SPIE, Vol. 6269, pp. 62695O-1–62695O-9. Barden, Samuel. 1998. "Review of Fiber-Optic Properties for Astronomical Spectroscopy." Fiber Optics in Astronomy III. Edited by Arribas, S., Mediavilla, E., and Watson, F. ASP Conference Series, Vol. 152, pp. 14–19. Barden, Samuel. 1994. "The Use and Benefits of Optical Fibers in Spectroscopy." Optical Astronomy from the Earth to the Moon. Astronomical Society of the Pacific Conference Series, Vol. 55, Proceedings of the Astronomical Society of the Pacific Summer Scientific meeting. Edited by Pyper, D. and R. Angione, pp. 130–138. Barnes, R., Raymond, S., Jackson, B., and Greenberg, R. 2008. "Tides and the Evolution of Planetary Habitability." Astrobiology, Vol. 8, Issue 3, pp. 557–568. 110 Baudrand, J., Guinouard, I., Jocou, L., Casse, M. 1998. "Use and Development of Fiber Optics on the VLT." Fiber Optics in Astronomy III. Edited by Arribas, S., Mediavilla, E., and Watson, F. ASP Conference Series, Vol. 152, pp. 32–43. Baudrand, J. and G. Walker. 2001. "Modal Noise in High-Resolution, Fiber-fed Spectra: A Study and Simple Cure." The Publications of the Astronomical Society of the Pacific, Vol. 113, Issue 785, pp 851–858. Bean, J., Seifahrt, A., Hartman, H., Nilsson, H., Wiedemann, G., Reiners, A., Dreizler, S., Henry, T. 2010. "The CRIRES Search for Planets Around the Lowest-Mass Stars. I. High-Precision Near-Infrared Radial Velocities with an Ammonia Gas Cell." The Astrophysical Journal, Vol. 713, Issue 1, pp. 410–422. Blake, C. H., Charbonneau D., White R. J., Marley M.S., Saumon D. 2007. "Multiepoch Radial Velocity Observations of L Dwarfs." The Astrophysical Journal, Vol. 666, Issue 2, pp. 1198–1204. Bland-Hawthorn, J. and P. Kern. 2009. "Astrophotonics: a new era for astronomical instruments." Optics Express, Vol. 17, Issue 3, pp. 1880–1884. Blankenship, D., Schmidt, B., Young, D., Schroeder, D., Greenbaum, J. 2011. "The Search for a Habitable Europa: Radar, Water and an Active Ice Shell." EPSCDPS Joint Meeting 2011, Vol. 6, pp. 1672–1673. Boisse, I., Bouchy, F., Chazelas, B., Perruchot, S., Pepe, F., Lovis, C., Hébard, G. 2011. "Consequences of spectrograph illumination for the accuracy of radialvelocimetry." Research, Science and Technology of Brown Dwarfs and Expolanets. Proceedings of an International Conference held in Shangai on 111 Occasion of a Total Eclipse of the Sun, Shangai, China. Edited by Martin, E.L., Ge, J., and Lin, W. EPJ Web of Conferences, Vol. 16, id. 02003. Bouchy, F. and F. Carrier. 2001. "P-mode oscillations on α Cen A.” Astronomy and Astrophysics, Vol. 374, p. L5–L8. Bouchy, A., Pepe, F., and Queloz, D. 2001. "Fundamental photon noise limit to radial velocity measurements." Icarus, Vol. 374, pp. 733-739. Braje, D., Kirchner, M., Osterman, S., Fortier, T., and Diddams, S. 2008. “Astronomical spectrograph calibration with broad-spectrum frequency combs.” The European Physical Journal D, Vol. 48, Issue 1, pp. 57–66. Brown, T., Gilliland, R., Noyes, R., and Ramsey, L. 1991. “Detection of Possible p-mode Oscillations on Procyon.” The Astrophysical Journal, Vol. 368, pp. 599–609. Buccino, A., Lemarchand, G., and Mauas, P. 2006. "Ultraviolet radiation constraints around the circumstellar habitable zones." Icarus, Vol. 183, Issue 2, pp 491–503. Butler, R., Marcy, G., Williams, E., McCarthy, C., Dosanjh, P., and Vogt, S. 1996. “Attaining Doppler Precision of 3 m/s.” Publications of the Astronomical Society of the Pacific, Vol. 108, pp. 500–509. Cassan, A., Kubas, D., Beaulieu, J., Dominik, M., Horne, K., Greenhill, J., Wambsganss, J., Menzies, J., Williams, A., Jorgensen, U., Udalski, A., Bennett, D., Albrow, M., Batista, V., Brillant, S., Caldwell, J., Cole, A., Coutures, C., Cook, K., Dieters, S., Prester, D., Donatowicz, J., Fouqué, P., Hill, K., Kains, N., Kane, S., Marquette, J., Martin, R., Pollard, K., Sahu, K., Vinter, C., Warren, D., Watson, B., Zub, M., Sumi, T., Szymanski, M., Kubiak, M., Poleski, R., Soszynski, I., Ulaczyk, K., Pietrzynski, G., Wyrzykowski, L. 2012. "One or more bound planets per Milky 112 Way star from microlensing observations." Nature, Vol. 481, Issue 7380, pp. 167–169. Casoli, F. and T. Encrenaz. 2007. The New World: Extrasolar Planets, Springer, New York, NY. Chazelas, B., Pepe, F., Wildi, F., Bouchy, F., Perruchot, S., and Avila, G. 2010. "New scramblers for precision radial velocity: square and octagonal fibers." Modern Technologies in Space- and Ground-based Telescopes and Instrumentation. Edited by Atad-Ettedguid, E. and D. Lemke. Proceedings of the SPIE, Vol. 7739, pp. 773947-1–773947-9. Chen, C.H., Reynolds, R., and Kost, A. 2006. "Origin of spectral modal noise in fibercoupled spectrographs." Applied Optics IP, Vol. 45, Issue 3, pp. 519–527. Clubb, Kelsey. 2008. "A Detailed Derivation of the Radial Velocity Equation." Department of Physics & Astronomy, San Francisco State University. < http://astro.berkeley.edu/~kclubb/pdf/RV_Derivation.pdf>. Corbett, J. and Allington-Smith, J. 2006. "Fibre modal noise issues in astronomical spectrophotometry." Ground-based Airborne Instrumentation for Astronomy. Edited by McLean, I. and M. Iye. Proceedings of the SPIE, Vol. 6269, pp. 62693N-1–62693N-11. Engle, S. G. and E. F. Guinan. 2011. "Red Dwarf Stars: Ages, Rotation, Magnetic Dynamo Activity and the Habitability of Hosted Planets." 9th Pacific Rim Conference on Stellar Astrophysics. Edited by Qian, S. ASP Conference Series. Engle, S. G., Guinan, E. F., and Mizusawa, T. 2009. "The Living with a Red Dwarf Program: Observing the Decline in dM Star FUV Emissions With Age." 113 FUTURE DIRECTIONS IN ULTRAVIOLET SPECTROSCOPY: A Conference Inspired by the Accomplishments of the Far Ultraviolet Spectroscopic Explorer Mission. American Institute of Physics Conference Proceedings, Vol. 1135, pp. 221–224. Figueira, P., Pepe, F., Melo, C., Santos, N., Lovis, C., Mayor, M., Queloz, D., Smette, A., Udry, S. 2010. "Radial Velocities with CRIRES. Pushing precision down to 5–10 m/s." Astronomy and Astrophysics, Vol. 511, id.A55. Fischer, Debra. "The Effects of Stellar Variability on the Detection and Characterization of Exoplanets." Astronomy of Exoplanets with Precise Radial Velocities. The Pennsylvania State University, August 2010. Web. 6 Feb. 2012. <http://exoplanets.astro.psu.edu/workshop/program.html>. Gatewood, G. and H. Eichhorn. 1973. "An unsuccessful search for a planetary companion of Barnard's star BD +4 3461." Astronomical Journal, Vol. 78, pp. 769–776. Geballe, T. "IR Transmission Spectra." Gemini Observatory. Apr. 2010. Web. 6 Feb. 2012. <http://www.gemini.edu/sciops/telescopes-and-sites/observing-condition- constraints/ir-transmission-spectra>. Gettel, S., Bender, C., Wright, J., and Wolszczan, A. 2011. "Measuring Radial Velocities in R-band with Telluric Line Calibration." American Astronomical Society, ESS meeting #3, #20.14. Gilliland, R. L. 1989. "Solar evolution." Global Planet, Vol. 1, pp. 35–55. Goodman, J.W. 1976. "Some fundamental properties of speckle." Journal of the Optical Society of America, Vol. 66, pp. 1145–1150. 114 Gough, D.O. 1981. "Solar interior structure and luminosity variations." Solar Physics, Vol. 74, pp. 21–34. Greenberg, R. 2005. Europa: The Ocean Moon: Search for an Alien Biosphere, Springer, New York, NY. Hart, M.H. 1978. "The evolution of the atmosphere of the Earth." Icarus, Vol. 33, pp. 23– 39. Hatzes, A., Cochran, W., and Endl, M. 2010. "The Detection of Extrasolar Planets: Using Precise Stellar Radial Velocities." Planets in Binary Star Systems. Astrophysics and Space Science Library, Vol. 366, pp. 51–75. Haynes, R., Bland-Hawthorn, J., Large, M., Klein, K., and Nelson, G. 2004. "New age fibers: the children of the photonic revolution." Optical Fabrication, Metrology, and Material Advancements for Telescopes. Edited by Atad-Ettedgui, E. and P. Dierickx. Proceedings of the SPIE, Vol. 5494, pp. 586–597. Haynes, D., Withford, M., Dawes, J., Lawrence, J., and Haynes, R. 2010. "Relative contributions of scattering, diffraction and modal diffusion to focal ratio degradation in optical fibres." Monthly Notices of the Royal Astronomical Society, Vol. 414, Issue 1, pp. 253–263. Heath, M., Doyle, L., Joshi, M., and Haberle, R. 1998. "Habitability of Planets Around Red Dwarf Stars." Origins of Life and Evolution of the Biosphere, Vol. 29, Issue 4, pp. 405–424. Hershey, J. L. 1973. "Astrometric analysis of the field of AC +65 6955 from plates tkane with the Sproul 24-inch refractor." Astrophysical Journal, Vol. 78, pp. 421–425. 115 Huang, S. S. 1959. "Occurrence of Life in the Universe." American. Scientist, Vol. 47, pp. 397–402. Jacob, W.S. 1855. "On certain Anomalies presented by the Binary Star 70 Ophiuchi." Monthly Notices of the Royal Astronomical Society, Vol. 15. pp. 228–238. Jones, H., Rayner, J., Ramsey, L., Henry, D., Dent, B., Montgomery, D., Vick, A., Ives, D., Egan, I., Lunney, D., Rees, P., Webster, A., Tinney, C., and Liu, M. 2008. "Precision Radial Velocity Spectrograph." Ground-based and Airborne Instrumentation for Astronomy II. Edited by McLean, I. and Casali, M. Proceedings of the SPIE, Vol. 7014, pp. 70140Y-1–70140Y-12. Joshi, M. 2003. "Climate Model Studies of Synchronously Rotating Planets." Astrobiology. Vol. 3, Issue 2, pp. 415–427. Joshi, M., Haberle, R., and Reynolds, R. 1997. "Simulations of the Atmospheres of Synchronously Rotating Terrestrial Planets Orbiting M Dwarfs: Conditions for Atmospheric Collapse and the Implications for Habitability." Icarus, Vol. 129, Issue 2, pp. 450–465. Kaltenegger, L. and W. Traub. 2009. "Transits of Earth-Like Planets." The Astrophysical Journal, Vol. 698, Issue 1, pp. 519–527. Kasting, James. 2011. "Habitable Planets: What Are We Learning from Kepler and Ground-Based Searches?" Astrobiology, Vol. 11, Issue 4, pp. 363–366. Kasting, James. 2010. How to Find a Habitable Planet, Princeton University Press, Princeton, NJ. Kasting, J. and D. Catling. 2003. "Evolution of a Habitable Planet." Annual Review of Astronomy and Astrophysics, Vol. 43, pp. 429–463. 116 Kasting, J., Whitmire, D., and Reynolds, R. 1993. "Habitable Zones around Main Sequence Stars." Icarus, Vol. 101, Issue 1, pp. 108–128. "Kepler Artwork." Kepler: A Search for Habitable Planets. NASA Ames Research Center. Web. 5 Feb. 2012. <http://kepler.nasa.gov/multimedia/artwork/>. "Kepler-22b, our first planet in the habitable zone of a Sun-like Star." Kepler: A Search for Habitable Planets. NASA Ames Research Center. 5 Dec. 2011. Web. 4 Feb. 2012. <http://kepler.nasa.gov/news/nasakeplernews/index.cfm?FuseAction=Show News&NewsID=165>. Khodachenko, M., Ribas, I., Lammer, H., Griessmeier, J., Leitner, M., Selsis, F., Eiroa, C., Hanslmeier, A., Biernat, H., Farrugia, C., and Rucker, H. 2007. "Coronal Mass Ejection (CME) Activity of Low Mass M Stars as an Important Factor for the Habitability of Terrestrial Exoplanets. I. CME Impact on Expected Magnetospheres of Earth-Like Exoplanets in Close-In Habitable Zones." Astrobiology, Vol. 7, Issue 1, pp. 167–184. Kjeldsen H., Bedding, T., Butler, R., Christensen-Dalsgaard, J., Kiss, L., McCarthy, C., Marcy, G., Tinney, C., and Wright, J. 2005. "Solar-like oscillations in α Centauri B." The Astrophysical Journal, Vol. 635, Issue 2, pp. 1281–1290. Lammer, H., Terada, N., Kulikov, Y., Lichtenegger, H., Khodachenko, M., and Penz, T. 2008. "Atmospheric Erosion Caused by Stellar Coronal Plasma Flows on Terrestrial Exoplanets within Close-In Habitable Zones of Low Mass Stars." 14th Cambridge Workshop on Cool Stars, Stellar Systems, and the Sun, ASP Conference Series, Vol. 384, pp. 303–312. 117 Laughlin, G., Bodenheimer, P., and Adams, F. 1997. "The End of the Main Sequence." The Astrophysical Journal, Vol. 482, pp. 420–432. Lawson, P., Traub, W., and Unwin, S. 2009. "Exoplanet Community Report." NASA, JPL Publication 09-3. Lemke, U., Corbett, J., Allington-Smith, J., and Murray, G. 2010. "Characterizing modal noise in fibre-coupled spectrographs for astronomy." Modern Technologies in Space- and Ground-based Telescopes and Instrumentation. Edited by AtadEttedgui, E. and D. Lemke. Proceedings of the SPIE, Vol. 7730, pp. 773924-1– 773924-13. Lemke, U., Corbett, J., Allington-Smith, J., and Murray, G. 2011. "Modal noise prediction in fibre-spectroscopy I: Visibility and the coherent model." Monthly Notices of the Royal Astronomical Society, Vol. 417, Issue 1, pp. 689–697. Li, C., Phillips, D., Glenday, A., Benedick, A., Chang, G., Chen, L., Cramer, C., Furesz, G., Kartner, F., Sasselov, D., Szentgyorgyi, A., and Walsworth, R. 2010. "Astrocomb calibration of an Echelle spectrograph." Ground-based and Airborne Instrumentation for Astronomy III. Edited by McLean, I., Ramsay, S., and Takami, H. Proceedings of the SPIE, Vol. 7735, pp. 77354O-1–77354O-8. Lo Curto, G., Lovis, C., Wilken, T., Avila, G., Chazelas, B., Esposito, M., Hansch, T., González-Hernández, J., Holzwarth, R., Ihle, G., Manescau, A., Pasquini, L, Pepe, F., Rebolo, R., Segovia, A., Sinclaire, P., Steinmetz, T., Udem, T., and Wildi, F. 2010. "Along the path towards extremely precise radial velocity measurements." Ground-based and Airborne Instrumentation for Astronomy III. 118 Edited by McLean, I., Ramsay, S., and Takami, H. Proceedings of the SPIE, Vol. 7735, pp. 77350Z-1–77350Z-8. Lovis, C. and Fischer, D. "Radial Velocity Techniques for Exoplanets." Exoplanets. Edited by Seager, S. University of Arizona Press, 2010. Lovis, C. and Pepe, F. 2007. "A new list of thorium and argon spectral lines in the visible," Astronomy and Astrophysics, Vol. 468, Issue 3, pp. 1115–1121. Luhman, K., Lada, C., Hartmann, L., Muench, A., Megeath, S., Allen, L., Myers, P., Muzerolle, J., Young, E., Fazio, G. 2005. "The Disk Fractions of Brown Dwarfs in IC 348 and Chamaeleon I." The Astrophysical Journal, Vol. 631, Issue 1, pp. L69–L72. Lunine, Jonathan. 2008. "Worlds Beyond: A Strategy for the Detection and Characterization of Exoplanets." Report of the ExoPlanet Task Force, Astronomy and Astrophysics Advisory Committee. Mahadevan, S., Ramsey, L., Redman, S., Bender, C., Terrien, R., Roy, A., Botzer, B., Osterman, S., Diddams, S., Ycas, G., and Quinlan, F. 2011. "Precision RVs in the NIR: First On-sky Velocities with a U/Ne Lamp and a Laser Frequency Comb." American Astronomical Society, ESS meeting #2, #2.06. Mahadevan, S., Ramsey, L., Redman, S., Zonak, S., Wright, J., Wolszczan, A., Endl, M., and Zhao, B. 2010. "The Habitable Zone Planet Finder Project: A Proposed High Resolution NIR Spectrograph for the Hobby-Eberly Telescope (HET) to Discover Low Mass Exoplanets around M Stars." Pathways Towards Habitable Planets. Edited by Coudé du Foresto, V., Gelino, D., and Ribas, I. Astronomical Society of the Pacific, pp. 272–277. 119 Martín, E. L., Guenther, E., Del Burgo, C., Zapatero Osorio, M., Deshpande, R., Rodler, F., Béjar, V., Esparza, P., Valdivielso, L., Moitinho, A., Amorim, A., Lima, J., Fiorenzano, A., Magazzu, A., and Tata, R. 2011. "High precision radial velocities in the near-infrared domain: Status and prospects." Research, Science and Technology of Brown Dwarfs and Exoplanets. Proceedings of an International Conference held in Shangai on Occasion of a Total Eclipse of the Sun, Shangai, China. Edited by Martín, E. L., Ge, J., and Lin, W. EPJ Web of Conferences, Vol. 16, id.02001. Martín E. L., Guenther, E., Zapatero Osorio, M., Buoy, H., and Wainscoat, R. 2006. "A Multiwavelength Radial Velocity Search for Planets around the Brown Dwarf LP 944-20." The Astrophysical Journal, Vol. 644, Issue 1, pp. L75–L78. Mayor, M., Bonfils, X., Forveille, T., Delfosse, X., Udry, S., Bertaux, J., Beust, H., Bouchy, F., Lovis, C., Pepe, F., Perrier, C., Queloz, D., and Santos, N. 2009. "The HARPS search for southern extra-solar planets. XVIII. An Earth-mass planet in the GJ 581 planetary system." Astronomy and Astrophysics, Vol. 507, Issue 1, pp. 487-494. Mayor, M. and D. Queloz. 1995. "A Jupiter-mass companion to a solar-type star." Nature, Vol. 378, Issue 6555, pp. 355–359. Mayor, M., Udry, S., Lovis, C., Pepe, F., Queloz, D., Benz, W., Bertaux, J., Bouchy, F., Mordasini, C., and Segransan, D. 2009. "The HARPS search for southern extrasolar planets. XIII. A planetary system with 3 super-Earths (4.2, 6.9, and 9.2 M )." Astronomy and Astrophysics, Vol. 493, Issue 2, pp. 639–644. 120 Moulton, F. R. 1899. "The limits of temporary stability of satellite motion, with an application to the question of the existence of an unseen body in the binary system F.70 Ophiuchi." The Astronomical Journal, Vol. 20, Issue 461, pp. 33–37. Murphy, M., Udem, T., Holzwarth, R., Sizmann, A., Pasquini, L., Araujo-Hauck, C., Dekker, H., D'Odorico, S., Fischer, M., Hansch, T., and Manescau, A. 2007. "High-precision wavelength calibration of astronomical spectrographs with laser frequency combs." Monthly Notices of the Royal Astronomical Society, Vol. 380, Issue 2, pp. 839–847. Newman, M.J. and R. T. Rood. 1977. "Implications of solar evolution for the Earth's early atmosphere." Science, Vol. 198, pp. 1035–1037. Noyes, R. W. and Weiss, N. O., Vaughan, A. H. 1984. "The Relation Between Stellar Rotation Rate and Activity Cycle Periods." The Astrophysical Journal, Vol. 287, pp. 769–773. Osterman, S., Diddams, S., Quinlan, F., Bally, J., Ge, J., and Ycas, G. 2011. "A Near Infrared Laser Frequency Comb for High-Precision Doppler Planet Surveys." Research, Science and Technology of Brown Dwarfs and Exoplanets. Proceedings of an International Conference held in Shangai on Occasion of a Total Eclipse of the Sun, Shangai, China. Edited by Martín, E. L., Ge, J., and Lin, W. EPJ Web of Conferences, Vol. 16, id.02002. Pogge, R. "Lecture 9: Stellar Spectra." The Ohio State University. 8 Jan. 2006. Web. 5 Feb. 2012. Quinlan, F., Ycas, G., Osterman, S., and Diddams, S. 2010. "A 12.5 GHz-Spaced Optical Frequency Comb Spanning >400 nm for Near-Infrared Astronomical 121 Spectrograph Calibration." Review of Scientific Instruments, Vol. 81, Issue 6, pp. 063105-1–063105-9. "The Radial Velocity Method." European Southern Observatory. 25 Apr. 2007. Web. 5 Feb. 2012. <http://www.eso.org/public/images/eso0722e>. Ramsey, L., Barnes, J., Redman, S., Jones, H., Wolszczan, A., Bongiorno, S., Engel, L., and Jenkins, J. 2008. "A Pathfinder Instrument for Precision Radial Velocities in the Near-Infrared." The Publications of the Astronomical Society of the Pacific, Vol. 120, Issue 870, pp. 887–894. Ramsey, L., Mahadevan, S., Redman, S., Bender, C., Roy, A., Zonak, S., Sigurdsson, S., and Wolszczan, A. 2010. "The Pathfinder Testbed: Exploring Techniques for Achieving Precision Radial Velocities in the Near-Infrared." Ground-based and Airborne Instrumentation for Astronomy III. Edited by McLean, I., Ramsay, S., Takami, H. Proceedings of the SPIE, Vol. 7735, pp. 773571-1–773571-6. Rawson, E., Norton, R., and Goodman, J. 1980. "Frequency dependence of modal noise in multimode optical fibers." Journal of Optical Society of America, Vol. 70, pp. 968–976. Raymond, S., Scalo, J., and Meadows, V. 2007. "A Decreased Probability of Habitable Planet Formation Around Low-Mass Stars." The Astrophysical Journal, Vol. 669, Issue 1, pp. 606–614. Redman, S. 2011. "Precise Radial Velocities in the Near Infrared." Ph.D. Dissertation, The Pennsylvania State University, University Park. Redman, S., Ycas, G., Terrien, R., Mahadevan, S., Ramsey, L., Bender, C., Osterman, S., Diddams, S., Quinlan,F., Lawler, J., and Nave, G. 2012. "A High-resolution Atlas 122 of Uranium-Neon in the H Band." The Astrophysical Journal Supplement, Vol. 199, Issue 1, article id. 2, pp. 1–11. Reid, I. N. and Hawley, S. L. 2005. New Light on Dark Stars: Red Dwarfs, Low-Mass Stars, Brown Dwarfs, 2nd ed., Springer. Rodler, F., Del Burgo, C., Witte, S., Helling, C., Hauschildt, P., Martín, E., Álvarez, C., and Deshpande, R. 2011. "Detecting planets around very cool dwarfs at near infrared wavelengths with the radial velocity technique." Astronomy and Astrophysics, Vol. 532, id.A31. Scalo, J., Kaltenegger, L, Segura, A., Fridlund, M., Ribas, I., Kulikov, Y., Grenfell, J., Rauer, H., Odert, P., Leitzinger, M., Selsis, F., Khodachenko, M., Eiroa, C., Kasting, J., and Lammer, H. 2007. "M Stars as Targets for Terrestrial Exoplanet Searches and Biosignature Detection." Astrobiology, Vol. 7, Issue 1, pp. 85–166. Segura, A., Kasting, J., Meadows, V., Cohen, M., Scalo, J., Crisp, D., Butler, R., and Tinetti, G. 2005. "Biosignatures from Earth-Like Planets Around M Dwarfs." Astrobiology, Vol. 5, Issue 6, pp. 706–725. Shapley, H. 1953. Climatic Change: Evidence, Causes, and Effects, Harvard University Press, Cambridge, MA. Solomonidou, A., Coustenis, A., Bampasidis, G., Kyriakopoulos, K., Moussas, X., Bratsolis, E., and Hirtzig, M. 2011. "Implications of possible internal liquid water oceans on Europa and other giant planets' satellites." EPSC-DPS Joint Meeting 2011. pp. 1218–1219. Steinmetz, T., Wilken, T., Araujo-Hauck, C., Holzwarth, R., Hansch, T., Pasquini, L., Manescau, A., D'Odorico, S., Murphy, M., Kentischer, T., Schmidt, W., and 123 Udem, T. 2008. "Laser Frequency Combs for Astronomical Observations." Science, Vol. 321, Issue 5894, pp. 1335–1352. Strughold, H. 1953. The Green and Red Planet, University of New Mexico Press, Albuquerque, NM. Tarter, J. 2007. "The evolution of life in the Universe: are we alone?" Highlights of Astronomy, Vol. 14, pp. 14–29. Tarter, J., Backus, P., Mancinelli, Ro., Aurnou, J., Backman, D., Basri, G., Boss, Al., Clarke, A., Deming, D., Doyle, L, Feigelson, E., Freund, F., Grinspoon, D., Haberle, R., Hauck, S., Heath, M., Henry, T., Hollingsworth, J., Joshi, M., Kilston, S., Liu, M., Meikle, E., Reid, I., Rothschild, L., Scalo, J., Segura, A., Tang, C., Tiedje, J., Turnbull, M., Walkowicz, L., Weber, A., and Young, R. 2007. "A Reappraisal of the Habitability of Planets Around M Dwarf Stars." Astrobiology, Vol. 7, Issue 1, pp. 30–65. Tofflemire, B., Wisniewski, J., Kowalski, A., Schmidt, S., Kundurthy, P., Hilton, E., Holtzman, J., and Hawley, S. 2012. "The Implications of M Dwarf Flares on the Detection and Characterization of Exoplanets at Infrared Wavelengths." The Astronomical Journal, Vol. 143, Issue 1, article id. 12. van de Kamp, P. 1963. "Astrometric study of Barnard's star from plates taken with the 24-inch Sproul refractor." The Astronomical Journal , Vol. 68, pp. 515–521. Walker, J. C. G. 1991. "Feedback Processes in the Biogeochemical Cycles of Carbon." Scientists on Gaia. Edited by Schneider, S. H. and P. J. Boston. MIT Press, Cambridge, MA. 124 West, A., Hawley, S., Bochanski, J., Covey, K., Reid, I., Dhital, S., Hilton, E., and Masuda, M. 2008. "Constraining the Age-Activity Relation for Cool Stars: The Sloan Digital Sky Survey Data Release 5 Low-Mass Star Spectroscopic Sample." The Astronomical Journal, Vol. 135, Issue 3, pp. 785–795. Wolszczan, A. and D. A. Frail. 1992. "A planetary system around the millisecond pulsar PSR1257 + 12." Nature, Vol. 355, pp. 145–147. Wood, T. 1984. "Actual modal power distributions in multimode optical fibers and their effect on modal noise." Optical Letters, Vol. 9, Issue 3, pp. 102–104. Ycas, G., Quinlan, F., Diddams, S., Osterman, S., Mahadevan, S., Redman, S., Terrien, R., Ramsey, L., Bender, C., Botzer, B., Sigurdsson, S. 2012. "Demonstration of On-Sky Calibration of Astronomical Spectra using a 2.5 GHz near-IR Laser Frequency Comb." Journal of the Optical Society of America. arXiv:1201.5125v1. Zugger, M., Kasting, J., Williams, D., Kane, T., and Philbrick, C. 2011. "Searching for Water Earths in the Near-infrared." The Astrophysical Journal, Vol. 739, Issue 1, article id. 12, pp. 1–5.