In the past few decades, our inventory of known planets has grown from just the eight in our solar system to over several thousand extrasolar planets detected around nearby stars through a variety of techniques (e.g. Mayor & Queloz 1995; Charbonneau et al. 2000; Marois et al. 2008). Extrasolar planets have been found orbiting the youngest stars to post-main sequence stars and around subsolar to intermediate-mass stars (Kraus & Ireland 2012; Currie et al. 2014b; Sato et al. 2003; Johnson et al. 2010). Statistical studies of exoplanets suggest that planet formation is common, occurring around about half of nearby Sun-like stars (e.g. Howard et al. 2013a and references therein).
Extrasolar planets are a diverse population, spanning a wide range in mass and orbital separation. Some exoplanets have similar properties (i.e. mass, orbital separation) to those in our own solar system. However, nearby exoplanets detected by two main methods – Doppler radial velocimetry and transit photometry, and distant exoplanets discovered by microlensing and direct detection, have revealed entirely new classes of planets such as hot Jupiters, Super-Earths, and wide-separation super-jovian planets.
Although extrasolar planets discovered thus far have provided important clues about the context within which the Earth and other solar system planets fit, our knowledge of the overall census remains highly incomplete. We have good constraints on the (high) frequency of planets larger than the Earth with orbital periods less than 100 days and on (the dearth of) the widest-separation, most massive planets (e.g. Howard et al. 2012; Nielsen et al. 2013). However, other classes of planetary systems are almost entirely unexplored – e.g. massive planets beyond a few AU, or planets with radii smaller than Earth. Furthermore, our knowledge about the physical properties of exoplanets (e.g. density, atmospheric/bulk composition) is much weaker, and is restricted to a handful of favorable cases. While we have detected both rocky planets and planets located on Earth-like orbits (e.g. Howard et al. 2013b), we have yet to conclusively identify a true (potentially habitable) Earth twin around a Sun-like star, let alone determine their frequency.
The Thirty Meter Telescope will provide an enormous advance in our ability to identify and characterize extrasolar planets. New technological advances -- e.g. high-precision Doppler measurements, high-precision space-based photometry, and advanced adaptive optics – have driven a large number of exoplanet discoveries. The TMT’s instrumentation will generate an incredible number of additional discoveries, will drastically expand the kinds of planets we can detect, will provide a rich understanding of these planets’ physical properties, and will potentially yield the first detections of habitable rocky planets.
Known extrasolar planets and Doppler sensitivity. Dashed lines show the typical detectable mass for a given level of Doppler precision, assuming edge-on circular orbits around a solar-type star.
With echelle spectrometers approaching the 0.1 m/s radial velocity (RV) measurement precision threshold needed to detect habitable Earth mass planets around Sun-like stars, the continuation of Doppler surveys on state-of-the-art telescopes is imperative. Such surveys, conducted with the optical and infrared echelle spectrometers being built for the TMT, have the potential to complete the census of Earth-mass planets in our stellar neighborhood. The collection of such systems will provide a valuable sample for follow-up programs to characterize their atmospheres through direct imaging.
Prior to the launch of the Kepler telescope in 2009, radial velocity surveys were the dominant method for discovering new exoplanetary systems, with over 400 systems discovered between 1995 and 2009 (http://exoplanets.eu). During the first decade and a half of the planet discovery era, Doppler surveys expanded their reach and progressed steadily from measurement precisions of 3-10 m/s to 1 m/s as a consequence of improvements in instrument stability and data analysis methods.
The radial velocity method originally focused on finding exoplanets around main sequence FGK stars. These targets are ideal, both due to their similarity to our Sun, and due to the abundance of atomic lines in their spectra. The gravitational velocity perturbation exhibited by the host star is directly related to the mass of the planet and is inversely related to its orbital distance. Hence, the RV detection method is most sensitive to massive, close-in planets. This ease of detection led to a substantial number of early discoveries of ‘hot Jupiters’, but as observing temporal baselines grew longer and as measurement sensitivities improved, the Doppler searches gradually expanded into more of the exoplanet phase space. As more planets were discovered, key physical trends began to emerge. Notable examples include a correlation between planet occurrence and stellar metallicity (Fischer & Valenti 2005; Johnson et al. 2010), a peak in the distribution of planets with orbits of 3 days (Cumming et al. 2008), and a gradual increase in the number of planets with increasing period and decreasing mass (Cumming et al. 2008). RV surveys have also lead to the discovery of complex planetary systems like HD 69830, which has three Neptune-mass planets and an asteroid belt (Lovis et al. 2006; Beichman et al. 2005), the multiple-planet system orbiting the M dwarf, GJ 581 (Mayor et al. 2009) and the planet around the nearby young dusty star ε Eridani (Hatzes et al. 2000; Benedict et al. 2006). RV planet detection programs are naturally limited to planets with periods that are shorter than the length of the survey, which currently amounts to somewhat more than 5 AU for the longest-running projects (Jupiter, for example has a 11.86 year orbital period and induces a 12 m/s radial velocity half amplitude). In recent years, a major focus has been on RV observations for characterization of transiting systems, particularly those discovered by the Kepler mission; the joint detection of a planet via radial velocimetry and transit photometry is exceptionally powerful, allowing measurements of planetary densities.
With multiple ongoing RV surveys focused on nearby stars, and improvements in both the hardware and RV data analysis techniques, imminent technical goals include pushing the RV precision beneath the ~1 m/s state-of-the-art threshold. Optical RV measurement precisions have advanced to the point where 1-2 m/s observations are routine with high temporal stability using high-resolution instruments such as Keck/HIRES Lick/APF, and Magellan/PFS (all using iodine cells) and ESO/HARPS and HARPS-N (using fiber scrambling and ThAr calibration). Under ideal circumstances, observers have reached 70 cm/s single measurement precision (Lovis et al. 2008). RV measurments have reached the point where they are confronted by limitations such as the intrinsic RV noise of F and G stars, and to where further advancement requires new technology for calibration sources (Lovis et al. 2006, Udry & Mayor 2008, Osterman et al. 2007, Li et al. 2008).
An RV precision goal of 0.1 m/s has been set for the ESPRESSO (Echelle SPectrograph for Rocky Exoplanet- and Stable Spectroscopic Observations, on-sky 2016) optical echelle spectrograph that will be deployed on the VLT (Pepe et al. 2013). Doppler precision measured in mere centimeters per second will only be accomplished with instruments that have very high spectral resolution (R>100,000) and fiber-scrambling capabilities to aid with spectral stabilization. Achieving this precision, of course, also requires very high signal-to-noise and hence a bright star with long exposures. Probing a large sample of stars at this precision will require a telescope such as TMT. The TMT partnership is developing MODHIS as one of the first light instruments, with a 0.1m/s RV precision goal in the near-IR. A concept for a high resolution optical spectrograph (HROS) with high radial velocity precision for a TMT 2nd generation instrument is being actively pursued.
The past decade has also shown the exceptional power of RV techniques in synergy with transit observations. Only for planets with both transit radius measurements and RV mass measurements can we determine density, and by extension, clues to the physical composition of the planets. With Kepler’s demonstration that small (1-4 RE) planets are common, the next step in understanding the frequency of earthlike planets is to determine the density of a significant number of these planets, especially those at large semi-major axes. The time needed for these RV observations are a significant bottleneck in completing Kepler’s scientific legacy. The TESS mission has a similar need for RV follow-up (albeit around brighter stars, but needing even higher levels of precision.) A high-precision RV capability, particularly optimized for late-type stars, on TMT, would be a powerful component of an integrated exoplanet roadmap, and would forge a near-complete understanding of nearby planetary systems.
A gallery of directly-imaged planets: (clockwise from the top-left) HR 8799 bcde (Marois et al. 2008, 2010a), Fomalhaut b (Kalas et al. 2008), GJ 504 b (Kuzuhara et al. 2013), HD 95086 b (Rameau et al. 2013), ROXs 42Bb (Currie et al. 2014b), and β Pic b (Lagrange et al. 2010). Other companions with masses below/near the deuterium-burning limit (~13 MJup) have also been imaged, many of them at wider separations (e.g., Chauvin et al. 2005; Goldman et al. 2010; Ireland et al. 2011; Luhman et al. 2011; Bowler et al. 2013; Delorme et al. 2013; Kraus et al. 2014; Bailey et al. 2014; Naud et al. 2014).
Direct imaging enables study of planets at wider separations and younger ages than are accessible to transit and radial-velocity methods. Observations obtained in the past 7 years on the largest ground-based telescopes equipped with adaptive optics (AO) have yielded the first images of planetary companions. The planets imaged thus far only trace the extremes of the population, typically orbiting at ~30–150 AU around stars more massive than the Sun and having masses of ~5–10 MJup. The vast majority of planets imaged to date have been detected from their thermal emission, with only one planet (or its surrounding disk) detected in reflected light (Fomalhaut b; Kalas et al. 2008) and two multi-planet systems imaged (HR 8799; Marois et al. 2008, TYC 8998-760-1; Bohn et al. 2020) so far. Imaged planets are mostly young (~1-100 Myr), where the planet-to-star contrasts are most favorable. Thus our current knowledge represents only a mere glimpse of the full range of planet demographics across stellar hosts, planet masses, orbital separations, and compositions.
Direct imaging discoveries thus far have sparked intensive follow-up efforts and, despite their limited number, illustrate the potential advances from comparative studies of exoplanets. These include the first insights into the atmospheric properties of young gas-giant planets. Near-infrared photometry indicates that young planets are dustier/cloudier than their older, field brown dwarf counterparts at similar effective temperatures (e.g., Bowler et al 2010a; Currie et al. 2011). Near-IR spectra clearly show the empirical hallmarks of reduced surface gravity, as expected given the young ages (e.g., Barman et al 2011; Currie et al 2014a; Bowler et al 2014). Such low gravities lead to thicker condensate clouds and enhanced non-equilibrium carbon chemistry in the young planetary photospheres compared to field brown dwarfs (e.g. Marley et al 2012; Skemer et al 2012, 2013). High-spectral-resolution data of imaged extrasolar planets have provided clues to planetary formation histories, by quantifying the relative abundances of carbon and oxygen in HR 8799c (Konopacky et al. 2013), yielding the rotation period of β Pic b (Snellen et al 2014), and highlighting ongoing accretion in the circumplanetary disk around GSC 06214-00210 (Bowler et al 2011). Spectra and photometry also show evidence for non-equlibrium chemistry. However, obtaining similar data for larger samples of gas-giant planets, as well as extending such work to lower-mass planets, is currently well out of reach.
Many more discoveries are being made with new planet-hunting instruments on 8-10 m telescopes such as the Gemini Planet Imager (GPI; Macintosh et al. 2014), the Spectro-Polarimetric High-contrast Exoplanet REsearch (SPHERE; Beuzit et al 2008), and the Subaru Coronagraphic Extreme Adaptive Optics project (SCExAO; Martinache and Guyon 2009). Large AO imaging surveys so far have shown that Jovian planets at large (tens of AU) separations are relatively rare (e.g. Vigan et al. 2012; Nielsen et al. 2013). On the other hand, radial-velocity data for 1.5–2 Msun stars show that Jovian planets increase in frequency with orbital separation out to several AU (e.g. Bowler et al. 2010b). Near-future AO surveys with these planet-hunting instruments should reveal much of the gas-giant population at intermediate separations. One of the most significant steps in the study of extrasolar planets in the years leading up to TMT will be direct imaging and spectroscopy of a large sample of young self-luminous exoplanets.
Direct imaging with the Thirty Meter Telescope will deepen our understanding of exoplanet properties and formation. Here we provide a brief list and and more details can be found in the TMT Detailed Science Case:
Predicted planet yields from TESS (Ricker et al. 2014). Both these surveys will be complete when TMT operations begin. TESS will find many planets around bright stars, while K2 will find larger numbers of planets around cooler stars.
With its unprecedented aperture and its access to about three-quarters of the observable sky, TMT will be well poised to carry out verification and follow-observations of transiting systems detected by other observatories. TMT's observations of transiting planetary systems will involve either spectroscopy or spectrophotometry to obtain more information about the properties of the planet or its host star, or high-spatial resolution imaging to identify stellar companions to stars or unassociated background objects. The observational modes will be dictated by the brightness of the target and the required cadence of the observations, e.g. at least several exposures per hour for monitoring during a planetary transit. Fainter stars or short cadence observations demand compensatory trade-off, i.e. with spectral resolution or resorting to spectrophotometry. This suggests that future instrumentation and configurations for TMT studies of transiting planets can be organized along three science goals: (1) characterization of faint host stars of transiting (candidate) planets using high-resolution spectroscopy and high angular-resolution imaging; (2) photometric and low-dispersion transit spectroscopy of the brightest systems observed by Kepler and K2, and many of the fainter systems discovered by TESS, (3) high-resolution transit spectroscopy of the brightest transiting systems discovered by TESS and eventually PLATO.
The NASA Kepler mission (2009-2013) dramatically expanded the scope of transit studies to Earth-size (and smaller) planets and to populations of thousands of planets on which robust statistical methods can be applied. However, Kepler surveyed only ~1/400th of the sky and the host stars of most Kepler systems are relatively distant (~1 kpc) and faint (V~14, K~12), making follow-up observations such as Doppler radial velocity measurements of mass or transit spectroscopy difficult or impossible even with 10-m class telescopes. For example, many Kepler host stars at ~1 kpc are probably unresolved binaries and it is often important to establish which star the planet transits to accurately determine its mass and rule out the possibility of a "false positive": the K-band diffraction limit of a 10-meter telescope is 55 AU and spectroscopic detection of single-line systems with separations of tens of AU requires high-SNR, high-resolution spectra that are difficult to obtain for these faint stars.
The next decade will welcome new space-based telescopes that will survey more of the sky and include many more brighter and nearby and hence observationally accessible host stars for transiting planets. The Kepler-2 or "K2" mission surveyed ~10,000 targets per field in 80-day campaigns (Howell et al. 2014, Prsa et al. 2014). The Transiting Exoplanet Survey Satellite (TESS) is surveying most of the sky (excepting the ecliptic plane) with durations of 27-351 d (Ricker et al. 2014). Both of these missions, especially TESS, is discovering transiting planets around bright stars, including a small number of Earth- to super-Earth-size planets around the brightest M dwarfs (see Figure 10.10). ESA's PLAnetary Transits and Oscillations of stars mission (PLATO) will launch in 2026, and is expected to discover hundreds of Earth- and super-Earth-size planets (as well as thousands of larger ones) around 4th-11th magnitude stars. PLATO will also discover numerous transiting planets in the habitable zones of solar-type stars (Rauer et al. 2014). Stars in this magnitude range offer the opportunity for high-cadence, high-spectral resolution observations of transits to search for emission associated with the planet, transit signals from other planets in the system, or physical properties of the stellar disk.
Thus TMT will have access to a census of thousands of comparatively poorly-studied Earth- and super-Earth-size Kepler and K2 planets and similar numbers of large and hotter planets around brighter and more accessible nearby stars, discovered by TESS and PLATO.
The high angular resolution of TMT IRIS AO imaging and IFU (Wright et al., 2010) could potentially play a significant role in characterizing the host stars for the ground-based microlensing planetary candidates in the next decade, and it would also aid space-based satellites such as WFIRST in studying the planet-host stars.
For the majority of microlensing planet detections, even though the precise planet-to-star mass ratios and angular planet-to-star separation can be directly inferred, the masses and distances of planet host stars cannot be separately constrained from the microlensing light curves alone. The degeneracy can be broken when additional microlens parallax signals are measured, which can be accomplished for a small fraction of exceptional events (e.g., Dong et al., 2009; Muraki et al., 2011) or by using a dedicated satellite to survey many events from another vantage point at d~AU separation from the Earth (Gould & Horne 2013; Dong et al., 2007).
An effective way to break the degeneracy is to directly measure the flux of the host star. For the ground-based observations, a microlensing target is typically blended with significant flux from unrelated stars due to crowding of the bulge field. High-resolution follow-up observations from Hubble Space Telescope (HST) or with AO systems on ground-based 10m-class telescopes have been performed to resolve the unrelated stars (e.g., Bennett et al., 2006, Dong et al., 2009, Janczak et al. 2010, Batista et al., 2014). The typical relative proper motion between the source and lens is ~6 mas/yr for a lens in the disk and ~4 mas/yr for a bulge lens, so it generally takes more than a decade for the lens to be separately resolved from the background star even with such high-resolution images (Han & Chang 2003). So far, the lens and source have been separately resolved for only two non-planetary microlensing events (Alcock et al., 2001, Kozlowski et al. 2007). Only the combined flux of the source and lens have been measured for planetary microlensing events with HST or AO, and the lens flux is constrained by subtracting the source flux, which can be precisely extracted from the light curve. In practice, the measurement uncertainty is dominated by the accuracy in aligning the high-resolution and microlensing light curve photometric systems, and the lenses have been robustly detected for source/lens contrast ratios up to factor of a few (Dong et al. 2009; Janczak et al. 2010).
With a 30m telescope, the lens would be separated from the source by ~2 λ/D in H-band in ~5 years after the peak of the detection of the event. Therefore, at the time of commissioning, TMT/IRIS would be able to resolve most of the planet hosts collected by a few years of operation by the KMTNet survey (see, e.g Henderson et al. 2015). An average source would be at H ~ 18, and for a typical ~0.3 Mʘ M dwarf lens at ~4 kpc, the required contrast ratio is about 10:1, easily observable by IRIS LGS. At ~2-3 λ/D, the achievable contrast ratio is ~1000:1, capable of detecting the lens population down to the bottom of the stellar mass function for the majority of events except those with the brightest giant sources, which only comprise a few percent of all KMTNet detections (Henderson et al. 2015). For some nearby lens systems, it is also possible to image the brown dwarf planet hosts. In these cases, the lens identification is considerably more secure, as its separation from the source can be checked with the independent relative source-lens proper motion due to finite-source effects, which are well measured for most planetary microlensing events. The lens flux measurements would offer a mass-distance relation for the lens after including models of extinction and mass-luminosity relation. The relative lens-source proper motion and event time scale provide tight constraints on the angular Einstein radius, which offer another mass-distance relation for the lens. The mass and distance of the lens star can be constrained down to ~10% by combining these constraints.
Once the lens is resolved, an exciting prospect for TMT is to obtain a medium resolution (R~4000) spectrum of the star with the IRIS near-IR IFU, which has the smallest plate scale of 9 mas. An exposure of approximately ten minutes on IRIS could obtain a spectrum of typical lens at reasonable SNR (≥10 per wavelength channel). The IRIS spectrum would enable identification of planet host’s stellar type, metal abundance as well as kinematic constraints through radial velocity measurements. It has been well established by RV surveys that the frequency of close-in Jovian planets increases as a function of host metallicity (Santos et al., 2004; Fischer & Valenti 2005). Recently, there is evidence from RV surveys and from the Kepler transiting planets that smaller planets at short-period orbits are distributed over a wide range of metallicity (Mayor et al., 2011, Buchhave et al., 2012). Spectroscopic characterization of microlensing hosts by TMT would open up a new window in studying whether the distribution of planets ranging from super Jupiters to sub-Earths at long periods depends on the environment of the star (metal rich vs. metal poor, bulge vs. disk).
During the operational span of a mission like WFIRST, the lens and source are typically not sufficiently separated to be completely resolved. However, thanks to a well-calibrated PSF, even when the lens and source are separated by a fraction of the PSF, it is possible to measure the lens flux from modeling the image elongation. It is expected that ~10% mass measurement can be routinely done in this way (Bennett et al, 2007; Spergel et al., 2013). TMT can still contribute to studying the WFIRST planet discoveries by making spectroscopic measurements of the planetary hosts. These measurements will determine important stellar parameters such as metallicity and surface gravity as well as imaging some low-mass hosts that are challenging for WFIRST.
