New studies and new worlds are challenging the core accretion theory, the primary model astronomers use to understand how planets form in our solar system and beyond.

For years, planetary scientists thought they had planet formation figured out. But as exoplanet discovery has exploded, we’ve found worlds around other stars that aren’t so easily explained. Now, two new studies presented at the winter meeting of the American Astronomical Society in Seattle are challenging the leading theory of planet formation, with varying degrees of success.

The Sub-Saturn Desert

Sub-Saturn (art)
An artist's illustration of the sub-Saturn planet designated OGLE-2012-BLG-0950
NASA / JPL / Goddard / F. Reddy / C. Ranc

According to core accretion theory, pebbles and rocks in the debris disk around an infant star will clump together into planet-sized cores. Near the star, there won’t be much material and only small, rocky planets will form. But farther from the star, much larger cores can come together. Once they become large enough (roughly ten times Earth’s mass), those cores quickly build up thick cloaks of gas. This process is what’s thought to have created the gas giants we see in the solar system today.

Runaway core accretion leads to planets of specific sizes: Either the gas dissipates before the runaway process begins at all, resulting in small, rocky worlds, or after the process has already taken off, resulting in gas giants. Either way, planets between the mass of Neptune and Saturn ought to be rare. Moreover, the cores that become big enough to set off runaway accretion mostly form beyond the “snow line,” where gases turn to ice and stick together to help planets grow.

Looking for sub-Saturns beyond the snowline — if they exist — isn’t easy. Most searches focus on planets close to their stars — without close-in orbits, astronomers won’t see enough transits or detect the stellar wobble that would confirm a planet’s existence.

But microlensing can find planets on far-out orbits. Microlensing happens when, from our vantage point, a star suddenly brightens as another star passes in front of it — the foreground star’s gravity bends and magnifies the background star’s light. If the foreground star hosts a planet, the planet will affect the magnification of background starlight too — even if it’s not super close to its star.

LCO / D. Bennett

Daisuke Suzuki (Institute of Space and Astronautical Science, Japan) and colleagues test the existence of the sub-Saturn desert using a group of 30 planets found by their microlensing effects. Turns out that there are ten times more planets in the Neptune-to-Saturn range than the group’s models of runaway core accretion would predict. In fact, the recent discovery of the microlens planet dubbed OGLE-2012-BLG-0950, also announced at the winter AAS meeting, fits right into the sub-Saturn desert. It has between 31 and 47 Earths’ worth of mass. (Find more details in the Keck Observatory’s press release.)

But core accretion is a complex theory. Suzuki’s group makes its predictions based on simulations of runaway core accretion that have lots of free parameters, like tuneable knobs on a stereo.

“In a year, I predict that the synthesis simulations will perfectly reproduce the observed distribution,” says Greg Laughlin (Yale University), who was not involved in the study. “This isn’t because they are correct, but rather because if you are allowed to vary a large number of parameters, any flexible framework should be capable of fitting any smoothly varying distribution of anything.”

In a nutshell, Suzuki and colleagues have thrown down the gauntlet at the feet of core accretion, but it remains to be seen whether the duel will proceed.

Wherefore Art Thou, Hot Jupiters?

A second study looks at the sub-Saturn desert from a different perspective. To make core accretion work for hot Jupiters — those gas giants that circle their stars on searingly close orbits — theorists must invoke migration. That is, if hot Jupiters first form farther out, then they must interact with the disk or gravitationally pinball off other planets to end up on their final, close-in orbit.

This core accretion-plus-migration scenario appears to jibe with mass-period diagrams that show a “desert” of hot Jupiters at super close-in orbits — presumably any hot Jupiters that move in any closer than that migrate right into their host star.

Planet-mass diagram
Planet mass-period diagrams reveal a "desert" of hot Jupiters on super close-in orbits. The gap could be a sign of Jupiters migrating inward from far-out orbits...
Elizabeth Bailey
In situ planet formation on the mass-period diagram
Elizabeth Bailey and Konstantin Batygin were able to postdict the sharp inner boundary on the mass-period diagram using in situ formation.
Elizabeth Bailey

But as Elizabeth Bailey (Caltech) pointed out at the AAS meeting, this “desert” has a sharply delineated inner boundary. If hot Jupiters were scattering in close to the star by gravitational interactions, this boundary ought to have a ragged appearance. Instead, Bailey finds that she can predict the exact shape of the boundary if planets come together right where they are, called in situ formation. “If this is correct, hot Jupiters are distinct in their origins from other cold Jupiters,” Bailey says.

In fact, that result may be just fine for core accretion, Laughlin says. Past studies have suggested that core accretion could proceed even in places with low density and high temperature, such as near the host star.

“[The study] shows that the curiously delineated distribution of hot Jupiters in the mass-period diagram is a natural consequence of in situ core accretion,” says Laughlin. “It does this, moreover, with no adjustable parameters.”

Indeed, Bailey and coauthor Konstantin Batygin (also at Caltech) emphasize that their result doesn’t rule out migration — migration might still happen in some systems and in fact probably did happen in the solar system — but it’s the exception rather than the rule.


D. Suzuki et al. “Microlensing Results Challenge the Core Accretion Runaway Growth Scenario for Gas Giants.Astrophysical Journal Letters, 2018 December 19. (Preprint available here)

A. Bhattacharya et al. “WFIRST Exoplanet Mass-measurement Method Finds a Planetary Mass of 39 ± 8 M_Earth for OGLE-2012-BLG-0950Lb.” Astronomical Journal, 2018 November 30. (Preprint available here)

E. Bailey & K. Batygin. “The Hot Jupiter Period–Mass Distribution as a Signature of in situ Formation.” Astrophysical Journal Letters, 2018 October 5. (Preprint available here)



Image of Rod


January 16, 2019 at 4:44 pm

Thanks for this report. Hot Jupiters, super-Earths, sub-Saturns, all missing from our solar system 🙂 I like the idea of in-situ formation. Planet building model(s) is not an easy task. "adjustable parameters" is always fun in computer models but could indicate that *stuff happens* too when explaining origins 🙂

You must be logged in to post a comment.

Image of stargate


January 19, 2019 at 10:43 pm

It is important to remember that multibody dynamical systems like a solar system under gravitational forces are nonlinear dynamical systems and hence are inherently chaotic. This means that over time their configurations or geometry will abruptly change due to gravitational perturbations in an unpredictable fashion. For example just look at the simple double pendulum.
What this means for planetary system models is that the orbital configurations observed today in all likelihood do not represent the initial configurations when the solar systems formed billions of years ago. Hence planetary formation models cannot in general be constrained by present day observations.
Closed-form deterministic solutions only exist for 2-body systems in a gravitational field. For more than 2 bodies the full nonlinear dynamical set of equations must be solved numerically. It is instructive to follow the evolution of such systems with computer models where the initial conditions are soon lost as the system evolves along a nonlinear landscape.

You must be logged in to post a comment.

Image of Rod


January 23, 2019 at 8:16 am

stargate, "Hence planetary formation models cannot in general be constrained by present day observations." This is an important point here concerning computer models. Free parameters in the simulation code and simulations explaining how planets evolved from dust grains and disks has much number crunching with various outputs in the programs. There will always exist in the simulations various objects formed at different stages in the evolution simulation - that a telescope will not see today, e.g. Jupiter migration in our solar system more than 4 billion years ago (Grand Tack) or the migration of various hot jupiters moving in very close to their host stars, stopping, and not falling into the star.

You must be logged in to post a comment.

You must be logged in to post a comment.