Why Are Most Known Exoplanets Hot Giants?

What the catalogue looks like

TerraForge carries a copy of NASA’s list of confirmed exoplanets, the planets found orbiting stars other than the Sun. At the time of writing it holds about 6,400 of them. If you scroll through it, a pattern jumps out quickly: an awful lot of these worlds are huge, and an awful lot of them are scorching hot.

The numbers bear that out. Of the planets with a known orbit, almost half go around their star in less than 10 days. More than four in five have a year shorter than Mercury’s, which is 88 days. About a third are at least six times wider than Earth, which puts them in Saturn and Jupiter territory, while fewer than one in ten are within a quarter of Earth’s size.

It would be easy to read this as a fact about the galaxy: most planets must be hot giants, and our solar system is the odd one out. That conclusion is wrong, or at least unsupported. The list is mostly a record of which planets our instruments find easiest to see. To understand why, you need to know how planets are found at all.

How planets are found

Almost no exoplanet has been seen directly. A planet is billions of times fainter than its star and sits right next to it in the sky, so it is lost in the glare. Instead, astronomers watch the star and look for the planet’s effect on it. Each technique is called a discovery method, and the archive records which one found each planet.

Transits

A transit happens when a planet passes in front of its star, as seen from Earth. For a few hours the star looks very slightly dimmer, then recovers. Space telescopes such as NASA’s Kepler and TESS stare at hundreds of thousands of stars and watch for those dips repeating at a regular interval. Roughly three planets in four on the list were found this way.

Radial velocity

A planet does not simply circle its star. Both bodies circle their shared centre of mass, so the star moves in a small loop too. When the star moves toward us its light is shifted slightly bluer, and when it moves away, slightly redder. That back-and-forth speed along our line of sight is called the radial velocity. The first planet found around a Sun-like star, 51 Pegasi b in 1995, was found this way.

The rest

Microlensing and direct imaging together account for only a few hundred planets, but they are worth knowing because their biases point the other way. Microlensing tends to catch planets around one to several AU from their star (an AU, or astronomical unit, is the distance from Earth to the Sun). Direct imaging works best on young giants far from their star, still glowing from the heat of their formation, like HR 8799 b.

Why big and close is easier

The two methods that found more than nine in ten known planets both reward the same thing: planets that are large, massive and close to their star.

Size sets the depth of the dip

How much a transit dims a star depends on how much of the star’s disc the planet covers. A Jupiter-sized planet in front of a Sun-like star blocks about 1% of its light. An Earth-sized one blocks less than a hundredth of a percent. The bigger dip is far easier to pick out from the star’s own natural flickering.

Closeness sets the odds and the repeats

A transit is only visible if the planet’s orbit happens to be lined up edge-on from our point of view. The closer the planet is to its star, the wider the range of viewing angles that still produces a transit. For a planet as far out as Earth, the chance is about half a percent. For a planet a twentieth of that distance, it is closer to one in ten.

Close planets also repeat faster. Astronomers usually want to see several evenly spaced dips before trusting a signal. A planet with a three-day year transits hundreds of times during a four-year survey like Kepler’s. A planet with a one-year orbit transits at most four times, and may be missed entirely if the telescope looks away. TESS watches most patches of sky for about 27 days at a time, which leans its finds even harder toward short orbits.

Mass and closeness set the wobble

The radial velocity method has the same leaning. A heavier planet pulls its star harder, and a closer planet pulls harder still and completes its loop sooner, so astronomers need fewer years of observation to see a full cycle. Jupiter moves the Sun by about 12 metres per second. Earth moves it by about 9 centimetres per second, far below what most instruments can reliably measure on most stars.

Put the two together and you get the planets that dominate the list: giants and mid-sized worlds on orbits of days or weeks. A planet that close to its star is heated hard, which is why so many of them are hot.

The 2014 and 2016 spikes

Count discoveries by year and the curve has two sharp spikes. In the catalogue, 2013 has about 130 new planets and 2015 about 150. In between, 2014 has nearly 870. In 2016 the count jumps to about 1,500, then drops back to around 150 in 2017.

Nothing special happened in the sky in those years. What changed was the bookkeeping. The Kepler telescope had already flagged thousands of candidate signals, dips that looked like planets but could also have been something else, such as a pair of stars eclipsing each other in the background. Checking each one by hand with follow-up observations is slow.

In 2014 the Kepler team confirmed 715 planets at once, using the fact that a star showing several separate transit signals is very unlikely to be fooled by background stars on all of them. In 2016 another analysis used statistics to rule out false alarms and confirmed 1,284 planets in a single announcement. Both batches came from the same telescope and carry its biases, which is why short orbits are so common in the list as a whole.

What this means in TerraForge

TerraForge builds a 3D world for every planet in the catalogue that has a measured mass, radius and orbit, and whose star has a measured temperature. Those four numbers are the minimum the engine needs to work out what a planet is made of and how warm it is. Just under 5,600 planets clear that bar, and they inherit every bias above.

The engine estimates each planet’s temperature from the star’s brightness and the planet’s distance, using its measured luminosity where the archive has one. The formula is the standard one for a planet’s equilibrium temperature: about 278 K for a body at Earth’s distance from a Sun-like star, rising with the fourth root of the starlight it receives. It ignores how much light the planet reflects, which is a deliberate simplification to match how the archive itself reports this value. The engine then adds warming from any greenhouse gases in the planet’s atmosphere. By that estimate, about three quarters of the planets TerraForge simulates are above 500 K, and well over a third are above 1,000 K.

That 1,000 K line matters, because it is where TerraForge starts calling a rocky world a lava world: hot enough for molten rock at the surface. Planets heavier than about 8 Earth masses never get there, because the engine assumes a core that heavy will have pulled in a thick envelope of gas while it formed, so it classifies them as some kind of giant. When it turns a catalogued planet into a recipe, anything at or above 50 Earth masses gets a gas-giant composition whatever its density says. Real planets near those lines are less tidy than this, which is a simplification worth keeping in mind.

Here is how the engine sorts the catalogue:

• Gas giants: about 2,000, or roughly 36%
• Lava worlds: about 1,000, or roughly 19%
• Barren rocky worlds: about 960, or roughly 17%
• Ice giants: about 860, or roughly 15%. TerraForge files hot Neptunes under this label too
• Rocky terrestrial worlds: about 460, or roughly 8%
• Brown dwarfs, water worlds, ice worlds and gas dwarfs: around 220 between them
• Earth-like worlds: 8

Planets such as KELT-9 b, a giant that circles a hot blue-white star in under a day and a half, and Kepler-10 b, a rocky world with a year of about 20 hours, are exactly what transit surveys find most easily. You can browse whole classes on the gas giant and lava world listings.

See it for yourself

Open the builderand load the Hot Jupiter preset. It puts a gas giant at 0.05 AU, a twentieth of Earth’s distance from the Sun. Look at its temperature and at the Closest Real Exoplanet card, which finds the catalogued planet most like the one you built. Then drag Distance from Star outward and watch the temperature fall as the starlight spreads thinner.

Next, build something like Earth: Initial Mass near 1, Star Type G, about 1 AU. That world would be one of the hardest of all for today’s telescopes to detect, with a tiny transit, a long year and a wobble too small to measure. Its scarcity in the catalogue tells you about our instruments, not about the galaxy.