Earth and Mars Were Built by Pebbles and Collisions, and Their Missing Volatiles Show It

A missing inventory written in stone

Earth and Mars are missing something. Compared with the Sun's raw ingredients, their bulk silicate—the rocky mantle and crust, excluding the metal core—is depleted in moderately volatile lithophile elements such as sodium, potassium, zinc, and indium. These are 'moderately volatile' because they condense at lower temperatures than the rock-forming elements magnesium and silicon, and 'lithophile' because they prefer silicate rock over metal. For decades, planetary scientists have asked what stripped them away.

Now a preprint posted to arXiv by Haiyang S. Wang and colleagues at the University of Copenhagen and other institutions argues that the answer is not one process but two. The depletion pattern, they suggest, is a chemical fingerprint of hybrid accretion: rocky planets growing both by pebble accretion—sweeping up millimeter- to centimeter-size dust clumps—and by collisional assembly from asteroid-scale planetesimals.

The puzzle has grown sharper as astronomers have revised their picture of the early solar system. In the traditional view, the inner disk was hot enough to keep volatile elements in the gas, preventing them from condensing into rocky building blocks. But in the emerging disk-wind paradigm, mass moves through the disk mainly via magnetic winds, and the midplane at Earth's orbit may have stayed below about 300 K. Samples returned from the asteroids Ryugu and Bennu also show that CI-chondrite-like material—primitive and solar-like except for the most volatile elements—was widespread. If the disk was not uniformly hot, another heat source must have driven volatile loss.

One candidate is the early planetesimals themselves. Short-lived radioactive isotopes or energetic collisions could have melted them, letting volatiles escape. Meteorites from Vesta record strong depletion of elements with sublimation temperatures below about 1200 K. The authors adopt Vesta as a stand-in for a population of early-formed, volatile-poor planetesimals.

The hot envelope that strips pebbles

The new ingredient in their model is the gas envelope of a growing protoplanet. As pebbles spiral inward, they pass through a hydrogen-helium atmosphere. The authors used a one-dimensional envelope model to track how hot and dense the gas becomes at the planet's surface. When the bottom temperature exceeds an element's sublimation temperature, that element turns to gas. Convection then lifts it upward, and recycling flows from the surrounding protoplanetary disk carry it away.

They tested this assumption with three-dimensional hydrodynamical simulations using the Athena++ code, following tracer particles in a convective envelope. The tracer particles were removed at the recycling boundary quickly: only about 30 percent remained after one year, and less than 3 percent after five years. Refractory elements such as magnesium, silicon, aluminum, and calcium are protected by a deep silicate vapor layer and an inner radiative zone, so they stay behind. The result is a depletion pattern that depends on each element's volatility.

What the numbers say for Earth—and for Mars

The team then asked what mixture of building blocks could reproduce the measured volatile depletion of bulk silicate Earth and bulk silicate Mars—their rocky mantles and crusts, again excluding the cores. They used 13 lithophile elements, from refractory aluminum to volatile indium, and a Bayesian inference framework—a statistical method that weighs competing models and estimates the probability of different mixtures.

For Earth, the best models favor a large contribution from two pebble-grown protoplanets: the proto-Earth and a giant impactor. A model with only those two bodies fits almost as well as one that adds Vesta-like planetesimals; the planetesimal component helps match lithium in particular. In the combined model, the median split is about 64 percent proto-Earth, 22 percent impactor, and 14 percent planetesimals. The authors summarize this as at least about 75 percent from pebble-grown protoplanets, with up to about 25 percent from Vesta-like planetesimals. If they allow the planetesimals to have a depletion curve unlike any known meteorite parent body, the planetesimal share can reach 40 percent, with an uncertainty range of roughly 15 percent. Models without a pebble-grown impactor, or without any pebble-grown body, are disfavored.

Mars looks different. Its best fit is 27 ± 5 percent pebble-accreted material and 73 ± 5 percent Vesta-like planetesimals. Tests that vary the assumed planetesimal depletion keep planetesimals dominant at over about 65 percent, while pebble accretion is never less than about 10 percent within 1σ uncertainty. The authors suggest Mars's pebble accretion was sluggish, perhaps because more massive protoplanets stirred its orbit.

Together, the results point to hybrid accretion. Earth experienced a relatively high pebble accretion rate, with a smaller but important contribution from already-depleted planetesimals. Mars's pebble accretion was comparatively sluggish, with planetesimal collisions dominating and only a modest share of pebbles.

The work does not claim to prove this scenario. It is a modeling study, not a direct observation of planet formation. The Vesta-like planetesimal composition is an assumption; the origin of Vesta's own volatile depletion is still debated. The Earth model assumes the Moon-forming giant impact did not remove moderately volatile elements, and collisional devolatilization remains poorly understood. The model is not yet precise enough to capture possible loss of silicon, magnesium, and chromium, and heavy halogens were excluded because their CI-chondrite abundances are disputed. The authors also tested pebbles that were already volatile-depleted before accretion; those models fit worse, though they cannot be ruled out.

Even with those caveats, the study shows how elemental abundances can become a record of growth. The same thermal processing that depleted volatile elements during Earth's and Mars's accretion would also affect rocky planets around other stars; life-essential volatiles such as carbon, hydrogen, nitrogen, oxygen, phosphorus, and sulfur are expected to be substantially depleted during the pebble accretion phase. Some could have arrived later, carried by volatile-rich pebbles, impacts, or primitive planetesimals that avoided heating.

Astronomers already see hints of volatile-poor rocky material in polluted white dwarf atmospheres and in stars that have ingested planetary bodies. If volatile depletion is a common fingerprint, then the chemical diversity of rocky exoplanets may partly reflect how they were assembled—through pebbles, through collisions, or through both.

Earth and Mars were not built by a single route. They were built through both drifting pebbles and the battered remnants of earlier worlds. Their missing volatiles are the ledger of that hybrid history.