Kepler-36 shows that planetary systems are less predictable than we thought
A collaboration with Brandon Weigel
In the early years of exoplanetology, astronomers knew about only one multi-planet system: the Solar System. They had simulations and models, of course, but throughout the 1990s, the theory behind those models was based mainly on our own planetary system. Even without any other data points, though, it seemed a reasonable assumption that most other exoplanetary systems were structured like our own: a set of terrestrial planets orbiting close to the host star, with giant planets similar to Jupiter and Saturn orbiting farther out.
This paradigm started to crumble in the mid-1990s, when the discovery of the hot Jupiter 51 Pegasi b turned conventional wisdom on its head. Massive gas giants simply shouldn’t orbit that close to their host stars! Slowly but surely, many of our assumptions about the structure of planetary systems have been shown to be completely false, as we discover counterexamples to long-held ideas about planet formation.
One of the most recent surprises orbits the subgiant star Kepler-36. It isn’t one planet, but two — exoplanets named Kepler-36b and Kepler-36c, with semi-major axes of 0.115 AU and 0.128 AU. This means that the two exoplanets are packed extremely close together. This in itself isn’t too peculiar; what is bizarre is that the two planets should be fairly alike, coming from the same area of the protoplanetary disk — but they aren’t. One is a dense, Earth-like terrestrial planet, while the other is a mini-Neptune with a gaseous envelope of hydrogen and helium.
So how do two completely different exoplanets form in essentially the same spot? That’s a good question — and its answer turns out to be crucial for our understanding of why exoplanets are so startlingly diverse. With Brandon Weigel, this week I’m doing some digging into why the planetary systems of the universe might be more varied than we thought.
Finding multiplanetary systems is hard!
Over the course of its nine-year mission, the Kepler space telescope monitored over half a million stars near the Sun. Kepler used the transit method for detecting exoplanets. It looked for small dips in the brightness of a star. If those dips recurred on a regular basis, it was strong evidence that they were caused by an exoplanet in orbit passing between Kepler and the star. Usually, combing data for an exoplanet candidate is as simple as looking for dips that have clear periods; you see each set gaps between transits.
For stars with multiple transiting exoplanets, however, things get tricky. These systems typically produce jumbled light curves that could easily be mistaken for other phenomena, like starspots — or the transits could be missed altogether. In the case of Kepler-36, there was an added problem. The two exoplanets are quite close to each other, and so they produce transit-timing variations, or TTVs — changes in the expected times of transits caused by their mutual gravitational pull.
Initially, the search algorithm used by Kepler completely missed Kepler-36b, which produced dips only about 17% as strong as those caused by Kepler-36c. A second algorithm, which takes potential TTVs into account, finally caught it, revealing a much richer system than astronomers originally thought (Carter et al. 2012). In fact, those TTVs, far from being a menace, ended up being a treasure trove of information. Typically, transits by a lone exoplanet only yield an estimate of its radius, but the TTVs allowed the team to modeling the gravitational forces between the planets for different trial masses — and hence derive their actual masses, which in turn provided a window into the exoplanets’ compositions.
The initial observations revealed masses of 4.45 and 8.08 Earth masses for Kepler-36b and Kepler-36c, respectively, and corresponding radii of 1.486 and 3.679 Earth radii. A simple calculation reveals densities of 7.46 grams per cubic centimeter — bit denser than Earth — and 0.89 grams per cubic centimeter, which is close to Saturn. The implications were clear: Kepler-36b is a rocky world with an iron-rich core, while Kepler-36c is rich in volatiles and holds onto an atmosphere composed mainly of hydrogen and helium.
This was a surprise. Despite orbiting only 0.01 AU from each other, the inner world was almost nine times as dense as its outer companion. Traditional models of planetary system formation predict that this sort of immense discrepancy should be impossible. The two exoplanets should be quite similar to each other. Yet the data told a different story.
A primordial solution for a primordial problem
The astronomers weren’t completely flabbergasted by the puzzle. Carter et al. briefly considered two possible solutions to the problem: migration or atmospheric erosion. The migration hypothesis, originally developed to explain the unexpected placement of hot Jupiters, suggests that exoplanets embedded in protoplanetary disks can move dramatically from the outer regions into close orbits around the star. This can be triggered by tidal interactions with the disk or perturbations with other planets. In this scenario, Kepler-36c would have formed far out, where it accreted volatiles and a substantial hydrogen/helium envelope, before being propelled into a tight orbit around its host star.
Lopez & Fortney 2013 were interested in exploring the second possibility. Protoplanets of all shapes and sizes may accrete large envelopes of hydrogen and helium during their early lives, but small, low-mass planets close to their host stars often lose these atmospheres, retaining heavy gases like oxygen and nitrogen. Extreme ultraviolet (XUV) radiation ionizes gas in the upper atmosphere and heats it up; this effect — called photoevaporation — is more pronounced on lighter molecules, like hydrogen and helium, and so bodies that experience high XUV fluxes tend to lose these gases fairly quickly.
Kepler-36b and Kepler-36c are quite close together, though, and if migration didn’t occur, they should have received the same amount of XUV flux. What, then, could cause one to lose most of its atmosphere? Lopez and Fortney suggested that one simple initial condition could have been different: core mass. It’s possible that Kepler-36b initially started out as a slightly less massive protoplanet than its neighbor, meaning it had a correspondingly lower escape velocity, and so it was easier for it to lose gas.
The theorists decided to test this out. They simulated a large set of exoplanet models, spanning a wide range of core masses and compositions. After simulating photoevaporative losses over the course of 7 billion years — the age of the system — they found parameters that reproduced the exoplanets’ derived properties. Kepler-36b started out with a core mass of 4.45 Earth masses — roughly the same as its present mass — and lost dramatic amounts of hydrogen and helium over the first 100 million years. After two billion years, its hydrogen/helium envelope was completely gone.
Kepler-36c, on the other hand, retained a significant amount of its envelope after starting with a core mass of 7.4 Earth masses. It also lost mass thanks to photoevaporation, but much more slowly, and not as dramatically. This let it end up as a Neptune-like object with a hydrogen/helium atmosphere, much different from its neighbor. Even if the two planets began with the same composition — 22% hydrogen and helium — the difference in core mass was enough to send them on two completely different paths.
What does this mean for exoplanetology?
The core mass hypothesis is extremely tempting. If true, it means that randomness in protoplanetary disks can naturally shape systems in many different ways. It removes the need for migration — a delicate process — to explain this sort of density discrepancy. Finally, it should be possible in any protoplanetary system — which is fortunate, since the same strange density contrast has since been observed in other pairs of exoplanets (see Kipping et al. 2014). At the moment, it may be a frontrunner to explain the Kepler-36 system.
Regardless of the mechanism behind this strange pair of exoplanets, they show that extremely diverse systems of exoplanets can exist. I don’t mean to say that any combination of masses, compositions and orbits can exist, but we should still expect to discover exotic systems that wouldn’t be out of place in, say, Star Wars. It wouldn’t be out of the question for a species living on a jungle world to hop on a spaceship and travel to a nearby small gas giant in a few months.
Still interested in what sort of exotic systems I’m talking about? Brandon Weigel wrote an awesome article about the exoplanets you might find — ocean worlds, iron planets, and much more. Check it out!
