When astronomers discovered the first exoplanet around a normal star 2 decades ago, there was joy—and bewilderment. The planet, 51 Pegasi b, was half as massive as Jupiter, but its 4-day orbit was impossibly close to the star, far smaller than the 88-day orbit of Mercury. Theorists who study planet formation could see no way for a planet that big to grow in such tight confines around a newborn star. It could have been a freak, but soon, more “hot Jupiters” turned up in planet searches, and they were joined by other oddities: planets in elongated and highly tilted orbits, even planets orbiting their stars “backward”—counter to the star’s rotation.
The planet hunt accelerated with the launch of NASA’s Kepler spacecraft in 2009, and the 2500 worlds it has discovered added statistical heft to the study of exoplanets—and yet more confusion. Kepler found that the most common type of planet in the galaxy is something between the size of Earth and Neptune—a “super-Earth,” which has no parallel in our solar system and was thought to be almost impossible to make. Now, ground-based telescopes are gathering light directly from exoplanets, rather than detecting their presence indirectly as Kepler does, and they, too, are turning up anomalies. They have found giant planets several times the mass of Jupiter, orbiting their star at more than twice the distance Neptune is from the sun—another region where theorists thought it was impossible to grow large planets. Other planetary systems looked nothing like our orderly solar system, challenging the well-worn theories that had been developed to explain it.
Nature is smarter than our theories.
“It’s been really obvious things didn’t fit pretty much from day one,” says Bruce Macintosh, a physicist at Stanford University in Palo Alto, California. “There has never been a moment when theory has caught up with observations.”
Theorists are trying to catch up—coming up with scenarios for growing previously forbidden kinds of planets, in places once thought off-limits. They are envisioning how planets could form in much more mobile and chaotic environments than they ever pictured before, where nascent planets drift from wide to narrow orbits or get ricocheted into elongated or off-kilter paths by other planets or passing stars. But the ever-expanding zoo of exotic planets that observers are tallying means every new model is provisional. “You can discover something new every day,” says astrophysicist Thomas Henning of the Max Planck Institute for Astronomy in Heidelberg, Germany. “It’s a Gold Rush situation.”
The traditional model of how stars and their planets form dates back to the 18th century, when scientists proposed that a slowly rotating cloud of dust and gas could collapse under its own gravity. Most of the material forms a ball that ignites into a star when its core gets dense and hot enough. Gravity and angular momentum herd the leftover material around the protostar into a flat disk. Dust is key to transforming this disk into a set of planets. The dust, which accounts for a small fraction of the disk’s mass, is made up of microscopic specks of iron and other solids. As they swirl in the roiling disk, the specks occasionally collide and stick together by electromagnetic forces. Over a few million years, the dust builds up into grains, pebbles, boulders, and, eventually, kilometer-wide planetesimals.
At that point gravity takes over, pulling in other planetesimals and vacuuming up dust and gas until planet-sized bodies take shape. By the time that happens in the inner part of the disk, most of its gas has been stripped away, either gobbled up by the star or blown away by its stellar wind. The dearth of gas means inner planets remain largely rocky, with thin atmospheres.
This growth process, known as core accretion, proceeds faster in the outer parts of the disk, where it is cold enough for water to freeze. The ice beyond this “snowline” supplements the dust, allowing protoplanets to consolidate more quickly. They build up a solid core five to 10 times the mass of Earth—quickly enough that the disk remains gas-rich and the core can pull in a thick atmosphere, producing a gas giant like Jupiter. (One of the goals of NASA’s Juno spacecraft, which arrived at Jupiter earlier this month, is to see whether the planet really does have a massive core.)
This scenario naturally produces a planetary system just like our own: small, rocky planets with thin atmospheres close to the star, a Jupiter-like gas giant just beyond the snowline, and the other giants getting progressively smaller at greater distances because they move more slowly through their orbits and take longer to hoover up material. All the planets remain roughly where they formed, in circular orbits in the same plane. Nice and tidy.
But the discovery of hot Jupiters suggested something was seriously amiss with the theory. A planet with an orbit measured in days travels an extremely short distance around the star, which limits the amount of material it can scoop up as it forms. It seemed inconceivable that a gas giant could have formed in such a location. The inevitable conclusion was that it must have formed farther out and moved in.
Theorists have come up with two possible mechanisms for shuffling the planetary deck. The first, known as migration, requires there to be plenty of material left in the disk after the giant planet has formed. The planet’s gravity distorts the disk, creating areas of higher density, which, in turn, exert a gravitational “drag” on the planet, causing it to gradually drift inward toward the star.
There is supporting evidence for the idea. Neighboring planets often end up in a stable, gravitational relationship known as orbital resonance. This happens when the lengths of their orbits are in a ratio of small whole numbers. Pluto, for example, orbits the sun two times for every three orbits of Neptune. It’s highly unlikely that they just happened to form that way, so they must have drifted into that position, where they were locked in by the extra stability. Migration early in our solar system’s history could account for other oddities, including the small size of Mars and the sparse, disrupted asteroid belt. To explain them, theorists have invoked a maneuver called the grand tack, in which Jupiter originally formed closer to the sun, drifted inward almost to the orbit of Earth, and then drifted out again to its current position.
Some modelers find such scenarios unnecessarily complex. “I do have faith in Occam’s razor,” says Greg Laughlin, an astronomer at the University of California (UC), Santa Cruz. Laughlin argues that planets are more likely to form in place and stay put. He says it’s possible for large planets to form close to their star if protoplanetary disks contain much more material there than previously believed. Some movement of planets may still occur—enough to explain resonances, for example—but “it’s a final subtle adjustment, not a major conveyor belt,” Laughlin says.
But others say that there simply could not be enough material to form close-in planets like 51 Pegasi b and others that are even closer. “They cannot have formed in situ,” physicist Joshua Winn of the Massachusetts Institute of Technology in Cambridge declares flatly. And the sizable fraction of exoplanets that appear to be in elongated, tilted, or even backward orbits also seems to imply some kind of planet shuffling.
For these oddballs, theorists invoke a gravitational melee rather than a sedate migration. A mass-rich disk could produce many planets close together, where gravitational tussles would fling them into the star, into weird orbits, or out of the system. Another potential disruptor is a companion star in an elongated orbit. Most of the time it would be too far away to have an influence, but occasionally it could swing in and stir things up. Or, if the parent star is a member of a tight-knit stellar cluster, a neighboring star might drift too close and wreak havoc. “There are a lot of ways to break a system,” Winn says.
Kepler's surprising finding that 60% of sunlike stars are orbited by a super-Earth, however, requires a whole new class of theories. Most super-Earths, thought to be largely solid rock and metal with modest amounts of gas, follow tighter orbits than Earth, and often a star has several. The Kepler-80 system, for example, has four super-Earths, all with orbits of 9 days or less. The traditional theory holds that inside the snowline core accretion is too slow to produce something so large. And super-Earths are rarely found in resonant orbits, suggesting that they haven’t migrated, but formed where they sit.
Researchers are coming up with ways around the problem. One idea is to speed up accretion, through a process known as pebble accretion. The gas in a rich disk exerts a lot of drag on pebble-sized objects. This generally slows them down, causing them to drift in toward the star. If they pass a planetesimal along the way, their slow speed means they can be captured more easily, boosting accretion. But faster accretion and a gas-rich disk raise their own problem: The super-Earths ought to pull in a thick atmosphere once they exceed a certain size. “How do you keep them from becoming gas giants?” asks astrophysicist Roman Rafikov of the Institute for Advanced Study in Princeton, New Jersey.
Eugene Chiang, an astronomer at UC Berkeley, says there is no need to speed up accretion, so long as the disk is solid-rich and gas-poor. He says that an inner disk 10 times denser than the one that formed the solar system could easily produce one or more super-Earths. Chiang has his super-Earths avoid collecting too much residual gas by forming in the dying days of the disk when most of the gas has dissipated.
Some early observations from the Atacama Large Millimeter/submillimeter Array (ALMA), an international facility nearing completion in northern Chile, support this proposal. ALMA can map radio emissions from the warm dust and gravel in disks. The few it has studied so far seem to be relatively massive. But the observations aren’t yet a smoking gun, because ALMA is not yet fully operational and it can only see the outer parts of disks, not the regions where super-Earths reside. “Getting close in, that’s the trick,” Chiang says—something that ALMA may perform when all 66 of its antennas are working.
Chiang also has an explanation for another discovery of Kepler’s: superpuffs, a rare and equally problematic set of planets that have a smaller mass than super-Earths but appear huge, with a puffed-up atmosphere making up 20% of their mass. Such planets are thought to form in a gas-rich disk. But in the inner disk, warm gas would fight against the planet’s weak gravity, so the cold and dense gas of the outer disk is the more likely womb. Chiang invokes migration to explain their close orbits—a notion supported by the fact that superpuffs are often found locked in resonant orbits.
Most of the attention in exoplanet research has so far focused on the inner parts of planetary systems, roughly within a distance equivalent to the orbit of Jupiter, for the simple reason that that’s all existing detection methods can see. The two main methods—measuring the wobble of stars caused by the gravitational tug of an orbiting planet and measuring the periodic dimming of a star as a planet passes in front—both favor big planets in close orbits. Imaging the planets themselves is extremely difficult, because their faint light is all but swamped by the glare from their star, which can be a billion times brighter.
But by stretching the limits of the world’s biggest telescopes, astronomers have seen a handful of planets directly. And over the past couple years, two new instruments designed specifically to image exoplanets have joined the hunt. Europe’s Spectro-Polarimetric High-contrast Exoplanet REsearch (SPHERE) and the U.S.-backed Gemini Planet Imager (GPI) are attached to big telescopes in Chile and employ sophisticated masks, called coronagraphs, to block out the light of the star. Not surprisingly, planets far from their stars are the easiest targets.
One of the earliest and most astounding systems found by direct imaging is the one around the star HR 8799, where four planets range in orbits from beyond that of Saturn out to more than twice the distance of Neptune. What’s most surprising is that all four are huge, more than five times the mass of Jupiter. According to theory, planets in such distant orbits move so slowly that they should grow at a glacial rate and top out at masses well short of Jupiter’s before the disk disperses. Yet the planets’ nice circular orbits suggest they weren’t flung there from closer to their stars.
Such distant giants lend support to the most radical challenge to standard theory, in which some planets form not by core accretion, but by a process called gravitational instability. This process requires a gas-rich protoplanetary disk, which breaks up into clumps under its own gravity. These blobs of gas would collapse over time directly into giant planets without having to form a solid core first. Models suggest that the mechanism will only work in particular circumstances: The gas has to be cold, it mustn’t be spinning too fast, and the contracting gas must be able to shed heat efficiently. Can it explain the planets of HR 8799? Only the outer two are distant and cold enough, Rafikov says. “It’s still quite a puzzling system,” he says.
In the past, radio telescope observations of protoplanetary disks have provided some support for gravitational instability. Sensitive to cold gas, the telescopes saw disks spattered with messy, asymmetrical blobs. But recent images from ALMA paint a different picture. ALMA is sensitive to shorter wavelengths that come from dust grains in the midplane of the disk, and its images of the star HL Tauri in 2014 and TW Hydrae this year showed smooth, symmetrical disks with dark circular “gaps” extending far beyond Neptune-like orbits (see picture below). “It was a tremendous surprise. The disk was not a mess, but has a nice, regular, beautiful structure,” Rafikov says. These images, suggestive of planets sweeping their orbits clean as they grow by core accretion, were a blow to advocates of gravitational instability.
It’s too early to tell what other surprises GPI and SPHERE may find in the outer reaches of planetary systems. But the region between those outlying neighborhoods and the close-in domains of hot Jupiters and super-Earths remains stubbornly out of reach: too close to the star for direct imaging, too far for indirect techniques relying on stellar wobbles or dimming. As a result, it is hard for theorists to get a full picture of what exoplanetary systems are like. “We’re basing things on fragmentary and incomplete observations,” Laughlin says. “Right now, everyone’s probably wrong.”
Astronomers won’t have to wait long for better data. Next year, NASA will launch its Transiting Exoplanet Survey Satellite (TESS), and the following year the European Space Agency (ESA) is expected to launch the Characterizing Exoplanets Satellite (CHEOPS). Unlike Kepler, which surveyed a large number of stars in sparse detail to compile an exoplanetary census, TESS and CHEOPS will focus on bright, sunlike stars close to Earth, enabling researchers to explore the midorbit terra incognita. And because the targeted stars are nearby, ground-based telescopes should be able to assess the mass of their planets, allowing researchers to calculate the planets’ density, indicating which are rocky or gassy.
The James Webb Space Telescope, due for launch in 2018, will go further, analyzing starlight that passes through an exoplanet’s atmosphere to determine its makeup. “Composition is an important clue to formation,” Macintosh says. For example, finding heavier elements in the atmospheres of super-Earths could suggest that a disk rich in such elements is needed to form planetary cores fast enough. And next decade, spacecraft such as NASA’s Wide Field Infrared Survey Telescope and ESA’s Planetary Transits and Oscillations will join the hunt, alongside a new generation of enormous ground-based telescopes with mirrors 30 meters across or more.
If the past is anything to go by, modelers will have to keep on their toes. “Nature is smarter than our theories,” Rafikov says.