More than 2000 planets have been discovered orbiting other stars. These extrasolar planets, or exoplanets, span a vast range of properties, and most form systems very different than our own, ranging from giant ‘hot Jupiters’ that are closer to their star than Mercury, to tightly-packed systems of multiple “super-earth” planets orbting faint red stars.
Planets are hundreds of thousand times to billions of times fainter than stars - nearly impossible to detect. The vast majority of these planets have been discovered through indirect techniques - changes in the parent star’s velocity or brightness caused by the presence of a planet. The exoplanet group at Stanford specializes in direct imaging of extrasolar planets, blocking out the light of the parent star to separate the planetary signal. With current technology such as the Gemini Planet Imager, this is only practical for massive young planets - the equivalent of our Jupiter, but only tens of millions of years old and shining brightly with heat released by its formation. Once these planets are detected, we can use spectroscopy to characterize them and determine their atmospheric composition and nature. Ultimately, the same technology will be applied to study Earthlike planets, allowing us to probe their atmospheres and hunt for compositions that could indicate life.
Gemini Planet Imager (GPI) observation of the young star 51 Eridani. After the light of the star has been blocked (indicated by the black mask), the signal from the planet (indicated by the arrow) is visible. GPI produces not just images but spectral cubes, measuring the spectrum of every pixel in the field of view. The spectrum of the planet can be extracted (right) showing absorption by methane and water vapor in its atmosphere
Gemini Planet Imager and instrumentation
The Gemini Planet Imager (GPI) is a revolutionary instrument designed to detect the faint light emanating from extrasolar planets. GPI is part of the modern instrument suite of the 8.1-meter Gemini South telescope located on Cerro Pachon, Chile. Lead by Professor Bruce Macintosh, the design and construction of GPI took place between 2008 and 2013; drawing from an international team of over 100 world renowned faculty, students, researchers, technicians and engineers.
GPI is designed to separate out and directly image the infrared light emitted from planets orbiting other stars. A planet seen directly can be characterized spectroscopically, allowing us to measure its atmospheric composition and to infer something of its nature and history. GPI utilizes thousands of tiny mirrors moving at 1000 times per second greatly reducing the blurring effect of the Earth’s atmosphere, resulting in higher-resolution images. A specialized coronograph blocks the light of the star, which would normally hide the much dimmer exoplanets, but allows the light from nearby objects to pass. Finally, an integral field spectrograph separates takes images and spectra of the area around the star looking for light from exoplanets. These spectra of exoplanets separate the light of the exoplanet according to wavelength, allowing astronomers to study the distant world’s composition, temperature, age and other characteristics, and better understand how planetary systems form and evolve. These instruments allow GPI to look for planets hidden by the bright glare of their parent stars, precisely track their position on the sky, and to understand their atmospheres.
Caption: Observing team at the Gemini telescope. From left to right: Ashley Chontos (summer student now at University of Hawaii), Vanessa Bailey (Now at JPL), Kate Follette (now at Amherst), Bruce Macintosh (Stanford) and Eric Nielsen (Stanford)
Properties and Circumstellar Environments of Young Stars
No matter the method used to study extrasolar planets, a guiding principle is that we only understand the planet as well as we understand the star. That’s true for direct imaging, where the age of the star is the key to converting the brightness of the planet we measure to the mass of the planet. Without an accurate age of the star, we cannot get an accurate planet mass. Most of the young stars close to the Sun reside in moving groups: groups of about 100 stars that were born together in the same molecular cloud, and though the cloud has since been blown away, the stars that formed inside are still moving in the same direction despite having spread out across the sky. These moving groups are known to have ages between 10 and 200 million years, but estimates of the ages of these groups vary from study to study. With many of the best GPIES targets, and all of the planets detected so far, residing in these moving groups it becomes of prime importance to ensure that we have robust age determinations for these groups.
In addition to searching for exoplanets orbiting young stars, our group also studies an even earlier period of solar system evolution---the protoplanetary disk stage. During the first 10 million years of a star’s lifetime, it is orbited by a circumstellar disk rich in molecular gas and “dust.” Direct studies of these systems provide the best opportunity to directly probe the sites of ongoing planet formation. With GPI, we can target these disks in infrared scattered light, to probe the distribution of micron-sized dust grains in the disk and uncover dynamically induced features like warps and spiral arms. Additionally, we use radio interferometers like the Atacama Large Millimeter/Submillimeter Array (ALMA) to target the thermal emission from millimeter-sized grains. In combination, these observations paint a detailed picture of the protoplanetary disk structure and present the “initial conditions” for theories of planet formation.
While Kepler has revolutionized our understanding of planets that are closer to their stars than the Earth is to the Sun, direct imaging is uniquely suited to probing giant planets at wider separations. An active area of research is measuring how frequently giant planets form at these larger orbital radii (at or beyond the giant planet region in our own solar system), and whether they form in the same way as close-in giant planets. When we look at other stars, do we see one population of giant planets at a large range of orbital periods, or are their multiple distributions, tracing different formation mechanisms or migration histories? Since the occurrence rate of these wide-separation giant planets is less than 10%, we need a direct imaging survey targeting hundreds of stars with high sensitivity to planets to answer this question.
The Gemini Planet Imager Exoplanet Survey (GPIES) was designed precisely to answer big questions about the formation of giant planet systems orbiting nearby stars. Targeting 600 young, nearby stars with an instrument optimized to detecting planets, after about 350 stars the survey has already detected six planets, including the first discovery of the planet 51 Eridani b. By combining these detections with a careful analysis of the overall sensitivity of the survey we measure a frequency of wide-separation giant planets around high-mass stars of about 7%. An exciting result is that this frequency of long-period giant planets is strongly dependent on the mass of the host stars: for high mass stars about 11% of stars have these planets, while for Sun-like stars and low-mass stars the frequency is less than 5%. This preference for giant planets to orbit high-mass planets is much stronger than what has been seen for giant planets at smaller separation. This suggests something qualitatively different happens when planets form around stars of different masses: either there are different formation mechanisms at play, or else the scales of the giant planet formation zone depends sensitively on the mass of the star.
Data analysis and exoplanet detection
The Gemini Planet Imager (GPI) is an incredible piece of technology that allows the detection of planets a million times fainter than their star and located at distances similar to the Gas Giant planets in our Solar System. The planet and its host-star are so close to each other that it is like seeing the firefly sitting on the side of the lense of a lighthouse from a kilometer distance.
Despite GPI’s state-of-the-art coronograph and adaptive optics, the glare of the star is still an order of magnitude brighter than the faintest planets we can detect. It is only by combining many images and applying advanced image processing techniques that we can achieve the required sensitivity to detect and characterize new planets.
Dozens of researchers and students have contributed for many years to the different pieces of the software pipeline. Our data architecture allows the GPIES’ team to automatically process images coming from the telescope enabling scientific discoveries even as the observations are still being made.
Our group at Stanford is working on applying more advanced statistical tools to improve our planet sensitivity and automate the detection process, therefore avoiding the long and fastidious task of checking the hundred of images by eye. Rigorous detection metrics are also essential to characterize the frequency of planets in the universe.