We were wrong: all stars don’t have planets, after all

It was only 30 years ago that humanity was discovering our first planets in orbit around stars other than our Sun. These first extra-solar planets, now known collectively as exoplanets, were unusual compared to the ones found in our own Solar System: they were Jupiter-sized, but located closer to their parent stars than Mercury is to our own. These “hot Jupiters” were just the tip of the iceberg, as they were merely the first that our detection technology became sensitive to.

The whole story changed a little over 10 years ago, with the launch of NASA’s Kepler mission. Designed to measure over 100,000 stars at once, simultaneously, by looking for a transit signal — where light from the parent star gets partially blocked, periodically, by an orbiting planet passing across its disk — Kepler discovered something astonishing. Based on the statistical likelihood of being serendipitously aligned with an orbiting planet’s geometry around its parent star, it averaged out so that practically all stars (between 80-100%) should possess planets.

Just a few months ago, we passed a milestone in exoplanet studies: more than 5000 confirmed exoplanets are now known. But surprisingly, a closer look at the known exoplanets reveals a fascinating fact: we may have vastly overestimated how many stars have planets, after all. Here’s the cosmic story of why.

In dense environments with many stars, such as young star clusters, the galactic center, or the centers of globular clusters, gravitational interactions could perturb the orbits of exoplanets, rendering them unstable. However, this may not be the explanation as to why no planets have been found in globular clusters; perhaps the metal-poor nature of the clusters examined is why no planets are present.

In theory, there are only two scenarios known that can form planets around stars. Both of them start off the same way: a molecular cloud of gas contracts and cools, and the initially overdense regions begin attracting more and more of the surrounding matter. Inevitably, whichever overdensity grows the most massive the most quickly begins to form a proto-star, and the environment around that proto-star forms what we call a circumstellar disk.

This disk will then develop gravitational imperfections within it, and those imperfections will attempt to grow via gravity, while forces from the surrounding material, the radiation and winds from the nearby stars and proto-stars, and interactions with other protoplanetesimals will work against their growth. The two ways that planets can then form, given these conditions, are as follows.

1. The core accretion scenario, where a sufficiently massive core of heavy elements — largely composed of rock and metal — can first form, with the remainder of a planet, including light elements and comet-like material, can accrete around it.
2. The disk instability scenario, where, far from the parent star, material rapidly cools and fragments, leading to quick collapse into a giant-sized planet.

According to simulations of protoplanetary disk formation, asymmetric clumps of matter contract all the way down in one dimension first, where they then start to spin. That “plane” is where the planets form, with that process repeating itself on smaller scales around giant planets: forming circumplanetary disks that lead to a lunar system. Superficially, these objects appear similar to some spiral galaxies.

Almost all of the planets that we discovered are only consistent with the core accretion scenario, but there were a few giant exoplanets, mostly discovered far from their parent star through direct imaging techniques, for which disk instability remained a strong possibility as far as how they were formed.

The disk instability scenario got a big boost in early 2022, when a team found a newly-forming exoplanet in a young protoplanetary system at a whopping three times the Sun-Neptune distance. Even better: they were able to see precisely at what wavelengths and where, relative to instabilities in the protoplanetary disk, the planet itself appeared.

This occurred at such a large radius from the parent star, and well beyond the radius at which core accretion processes can explain the formation of such a massive planet so early in a stellar system’s life cycle, that it could only have formed via the disk instability scenario. We now believe that the overwhelming majority of gas giant planets formed at extremely large distances from their parent stars likely formed via the disk instability scenario, while the closer-in planets must have formed via the core accretion scenario.

A dusty disk of protoplanetary material (red) surrounds the inner stellar system (blue) around the young star AB Aurigae (yellow star), with a candidate planet revealed in the location identified by the green arrow. This object has properties that render it incompatible with the standard core accretion scenario.

It’s only because of what we’re most sensitive to — large changes in either the parent star’s apparent motion or apparent brightness over short timescales — that the majority of planets we’ve found must have formed via core accretion. The reality is that we don’t have sufficient data to identify the overwhelming majority of Jupiter-sized planets at very large distances from their parent stars. This may be something, given the coronagraphic capabilities of new observatories like JWST and the presently under-construction thirty meter-class ground-based telescopes here on Earth, that gets remedied over the coming years.

The disk instability scenario doesn’t have any dependence on how many heavy elements are available to form rock-and-metal cores for planets, so we can fully expect, at very large distances from a star, to find the same number of planets regardless of what abundance of heavy elements exist in that particular stellar system.

But for the core accretion scenario, which ought to apply to all planets found with orbital periods ranging from hours to a few Earth-years, there should be a limit. Only stars with circumstellar disks that possess at least a critical threshold of heavy elements should be able to form planets via core accretion at all.

The mass, period, and discovery/measurement method used to determine the properties of the first 5000+ (technically, 5005) exoplanets ever discovered. Although there are planets of all sizes and periods, we are presently biased toward larger, heavier planets that orbit smaller stars at shorter orbital distances. The outer planets in most stellar systems remain largely undiscovered, but those that have been discovered, largely through direct imaging, are difficult to explain the way we think most exoplanets form: via the core accretion scenario.

This is a wild realization with far-reaching implications. When the Universe began some 13.8 billion years ago with the onset of the hot Big Bang, it rapidly formed the earliest atomic nuclei through nuclear fusion processes that occurred during those first 3-4 minutes. Over the next few hundred thousand years, it was still too hot to form neutral atoms, but too cold for any further nuclear fusion reactions to occur. Radioactive decays could still occur, however, bringing an end to any unstable isotopes that existed, including all the Universe’s tritium and beryllium.

When neutral atoms first formed, we then possessed a Universe consisting of, by mass:

• 75% hydrogen,
• 25% helium-4,
• ~0.01% deuterium (a stable, heavy isotope of hydrogen),
• ~0.01% helium-3 (a stable, light isotope of helium),
• and ~0.0000001% lithium-7.

That last component — the tiny amount of lithium in the Universe — is the only element that falls into the “rock and metal” category. With only one-part-in-a-billion of the Universe made of something other than hydrogen or helium, we can be confident that the very first stars of all, made out of this pristine material left over from the Big Bang, could not have formed any planets via core accretion.

A sample of 20 protoplanetary disks around young, infant stars, as measured by the Disk Substructures at High Angular Resolution Project: DSHARP. Observations such as these taught us that protoplanetary disks form primarily in a single plane and tend to support the core accretion scenario of planet formation. The disk structures are seen in both infrared and millimeter/sub-millimeter wavelengths.

That means that rocky planets simply weren’t possible in the earliest stages of the Universe!

That simple but essential realization, in itself, is revolutionary. It tells us that there must be a minimal amount of heavy elements created in the Universe before planets, moons, or even giant planets in close proximity to their parent stars can exist. If planets and/or other rocky worlds are required for life, a plausible but uncertain conjecture, then life could not have come into existence in the Universe until enough heavy elements existed to form planets.

This was bolstered in the 2000s, when two large studies were done searching for stars with transiting planets within the two brightest globular clusters as seen from Earth: 47 Tucanae and Omega Centauri. Despite having at least hundreds of thousands of stars inside, no planets were ever found around any of them. One possible reason put forth was that, with so many stars in such a densely packed region of space, perhaps any planets would be gravitationally ejected from their stellar systems. But there’s another reason that must be considered in this new context: perhaps there simply weren’t enough heavy elements present in these ancient systems to form planets back when the stars formed.

In fact, that’s a very compelling explanation. The stars in 47 Tucanae largely formed all at once some ~13.06 billion years ago. An analysis of the red giant stars inside revealed that they contain only about 16% of the heavy elements found in the Sun, which may not be enough to form planets via core accretion. Omega Centauri, by contrast, had multiple periods of star formation inside, with the most heavy element-poor stars having just ~0.5% of the heavy elements that the Sun possesses, while the most heavy element-rich stars have about ~25% of the heavy elements present in the Sun.

You might then think to look at the largest data set we have — the full suite of all 5069 (at the present moment) confirmed exoplanets — and ask, of the exoplanets found with orbital periods under ~2000 days (about 6 Earth years), how many of them are known with extremely low contents of heavy elements?

• Only 10 exoplanets orbit stars with 10% or fewer of the heavy elements found in the Sun.
• Only 32 exoplanets orbit stars with between 10% and 16% of the Sun’s heavy elements.
• And only 50 exoplanets orbit stars with between 16% and 25% of the Sun’s heavy elements.

That means, all told, that only 92 out of 5069 exoplanets — just 1.8% — exist around stars with a quarter or fewer of the heavy elements found in the Sun.

This diagram shows the discovery of the first 5000+ exoplanets we know of and where they’re located on the sky. Circles show location and size of orbit, while their color indicates the detection method. Note that the clustering features are dependent on where we’ve been looking, not necessarily on where planets are preferentially found. No planets have been found within globular clusters, including the long-imaged 47 Tucanae and Omega Centauri.

There’s one exoplanet around a star with fewer than 1% of the Sun’s heavy elements (Kepler-1071b), a second around a star with about ~2% of the Sun’s heavy elements (Kepler-749b), four of them around a star with about 4% of the Sun’s heavy elements (Kepler-1593b, 636b, 1178b, and 662b), and then four additional ones with between 8-10% of the Sun’s heavy elements.

Travel the Universe with astrophysicist Ethan Siegel. Subscribers will get the newsletter every Saturday. All aboard!

In other words, when we look at the exoplanets that exist around stars in detail, we find that there’s a steep drop-off in their abundance based on how many heavy elements are present. Below about 20-30% the heavy element abundance of the Sun, there’s a “cliff” in the exoplanet population, with an extremely steep decline in exoplanet abundance altogether.

Based on what we know about heavy elements and how/where they form, this holds a significant set of implications for the chances of rocky planets and moons — and hence, for living, inhabited worlds — all across the Universe.

The very first stars to form in the Universe were different than the stars today: metal-free, extremely massive, and destined for a supernova surrounded by a cocoon of gas. Stars require a series of cosmic steps to occur before they can form, and cooling the neutral, pristine matter is a key and critical step.

The very first stars that form are the first stars to produce heavy elements like carbon, oxygen, nitrogen, neon, magnesium, silicon, sulphur, and iron: the most abundant elements in the Universe other than hydrogen and helium. But they’re only capable of increasing the heavy element abundance up to about ~0.001% of what we find in the Sun; the next generation of stars to form will remain exceedingly poor in heavy elements even though their contents are no longer pristine.

This means that many generations of stars, all processing, re-processing, and recycling the detritus from each prior generation, must exist in order to build up enough heavy elements to form a rock-and-metal-rich planet. Until a critical threshold of those heavy elements is met, Earth-like planets are impossible.

• There will be a period of time, lasting more than half a billion years and perhaps more than a full billion years, where no Earth-like planets can form at all.
• There will then come a period, lasting several billion years, where only the richest, central regions of galaxies can possess Earth-like planets.
• After that, there will be another period of several billion years where the central galactic regions and portions of the galactic disk can possess Earth-like planets.
• And then, up to and including the present day, there will be many regions, particularly in the outskirts of galaxies, in the galactic halo, and in globular clusters found throughout the galaxy, where heavy element-poor regions still cannot form Earth-like planets.

This color-coded map shows the heavy element abundances of more than 6 million stars within the Milky Way. Stars in red, orange, and yellow are all rich enough in heavy elements that they should have planets; green and cyan-coded stars should only rarely have planets, and stars coded blue or violet should have absolutely no planets at all around them. Note that the central plane of the galactic disk, extending all the way into the galactic core, has the potential for habitable, rocky planets.

When we looked only at the raw numbers and extrapolated based on what we had seen, we learned that there are at least as many planets as there are stars in the Universe. This remains a true statement, but it’s no longer a smart bet to presume that all, or almost all, of the stars in the Universe possess planets. Instead, it looks like planets are most abundant where the heavy elements that are needed to form them via core accretion are also most abundant, and that the number of planets that exists drops off as their parent stars possess fewer and fewer elements.

The drop-off is relatively slow and steady until you reach somewhere around 20-30% the abundance of elements found in the Sun, and then there’s a cliff: a steep drop-off. Below a certain threshold, there should be no planets that form via core accretion — including all potential Earth-like planets — at all. It took billions of years before most newborn stars would have planets around them, and has severe implications that restrict the possibilities for life in globular clusters, the outskirts of galaxies, and all throughout the Universe at early cosmic times.

Today’s Universe may be teeming with planets, and perhaps with inhabited planets as well, but this hasn’t always been the case. Early on, and anywhere the heavy element abundance remains low, the needed ingredients simply weren’t around.