A long popular-science piece about space came up in my feed—the kind that runs from the Big Bang to extraterrestrial life in one sitting. Somewhere along the way, it said this: our Solar System is suspiciously rich in metals, significantly richer than the average Milky Way system of the same age, and no one has clearly explained why. The author tossed the line in as scenery and moved on, to Europa and Titan.
The line stuck. Because “no one has explained it” in astrophysics almost always means not a mystery, but a two-layered story: beneath the question “why is it so” lies the question “is it really so?” The Sun does indeed have a measured anomaly, with a name of its own, fifty-odd papers, and solid numbers. But it is phrased nothing like the popular retelling, and over the past two years the whole matter has taken a fairly rough turn. The topic has not been covered in the archive: neither “refractories,” nor the “protosolar cloud,” nor the helioseismic crisis have come up here before.
First, it is worth separating two different claims that easily fuse into a single line about an “anomalous system.” The first is about metallicity. By itself, that means nothing: the Sun is the benchmark here; all stellar abundances are measured against it. What matters is how the Sun looks beside its neighbors of the same age and from the same part of the Galaxy.
And here the numbers are modest: in the most complete and least biased sample of stars within 25 parsecs, the median metallicity is [Fe/H] = −0.08, while the Sun is richer than roughly 65 % of these stars. In another study, the sample was deliberately narrowed to the Sun’s contemporaries at the same Galactic radius—the average came out to −0.09, and the Sun outweighs it by a tenth of a dex. This is a statistical shift, not a geochemical anomaly of the neighborhood: the scatter within the sample is twice as large as the shift itself.
The second claim is the one the literature calls the Meléndez effect: the Sun is relatively poorer in refractory elements than the overwhelming majority of stars that resemble it like two drops of water. Refractories are elements with high condensation temperatures, the ones that first precipitate into dust as gas cools: aluminum, calcium, titanium, silicon, iron, magnesium. Volatiles—carbon, nitrogen, oxygen, sulfur—remain gaseous longer. Comparing stars by the abundance of each individual element is pointless; what gets compared is the slope: how the difference in abundances relative to the Sun changes with an element’s condensation temperature. The Sun’s slope is negative, and this is no triviality—the effect has an amplitude of about 20 %, a perfectly measurable quantity.
Measuring something like this is technically unpleasant, and the workaround deserves a paragraph of its own. You cannot put the Sun in a telescope beside a stellar twin and take two spectra with the same instrument—the difference in energy flux at Earth between the two objects is enormous, and the systematics of the analysis would eat the entire signal. So reflected sunlight from an asteroid or the galactic background is used as the reference, recorded with the same spectrograph. The error arising from the fact that the Sun is seen almost from its equatorial plane while a star is seen at a random angle to its rotational axis has been checked separately: observations of the solar disk at different latitudes yield discrepancies no greater than 0.005 dex.
The first attempt, in 2009—Jorge Meléndez and co-authors, eleven carefully selected twins (temperature, surface gravity, and metallicity within narrow ranges, spectra at a resolution of 65,000 and a signal-to-noise ratio of about 450), and an analysis strictly line by line, element by element, relative to the Sun. The result: the abundance difference rises linearly with condensation temperature, the full range is about 0.08 dex, and only one twin out of eleven had a profile resembling the Sun’s. It looked like a discovery: something had been taken from the Sun.
Nine years later, Megan Bedell and co-authors repeated the experiment with 79 stars within 100 parsecs, at still higher resolution and with errors in the hundredths of a dex, and confirmed it: 80 % of stars of this kind are enriched in refractories relative to the Sun. Since then, the number “80 %” has migrated through every article as a point of reference. A 2025 review estimates that the effect, in one form or another, affects 4–7 % of solar-type stars—that is, the phenomenon is not unique, but neither is it universal.
Then comes the most beautiful part—the hypotheses explaining precisely where the refractories disappeared to from the Sun’s photosphere.
Version one, from that same 2009 paper: the planets took them. If you mentally stir Mercury, Venus, Earth, and Mars into the Sun’s convective zone, the resulting spectrum becomes much closer to those of the twins. The logic is appealing: the Sun has rocky planets, therefore it has a “warehouse,” and the meager composition of the photosphere is simply a receipt showing that the material never fell back onto the star.
Version two, from Ryan Booth and James Owen, is even more elegant. Jupiter, having formed early, cuts an annular gap in the protoplanetary disk. At its outer edge, a dust trap arises: gas pressure is elevated there, and solid particles are slowed down. Gas flows calmly through the barrier and keeps falling onto the star, while dust carrying refractories does not. The star grows, but it grows from depleted material. Hydrodynamic modeling in 2023 showed that with a fortunate choice of parameters, this mechanism produces roughly half the observed amplitude; another paper that same year added a counterargument—particles in the trap collide and shatter quickly, and the fragments seep through the barrier without trouble. So the “leaky trap” remains an open question.
But both planetary versions have a common enemy, and it is purely an engineering one—chronology. The protoplanetary disk disperses in less than 10 million years, while the young Sun’s deep convective zone needs more than 25 million years to become thin (today it contains only a few percent of the star’s mass). If the material fell onto the star before the zone became thin, its signature would simply drown there and never reach the photosphere. A workaround was proposed through episodic accretion: under it, the stellar model quickly jumps off the Hayashi track and reaches the main sequence with a thin convective zone in just a few million years. Worse still, a planetary signature has an expiration date: an excess of heavy elements sinks below the convective zone through thermohaline, “fingering” instability. Swallowing a planet erases its own tracks.
The planetary versions make a testable prediction: if planets took the refractories, then planet-hosting stars should be chemically closer to the Sun than stars without planets. In 2024–2025, this prediction was killed three times, with different instruments and different samples. A study using Gaia data on 17,412 solar analogs (medium-resolution spectra, abundances reconstructed by a trained model, precision of 0.04–0.1 dex; the sample included 50 stars with planets and 61 planets) produced a flat result: the Sun remains depleted regardless of what planets a star has, and there is no difference between hosts of hot giants, ordinary giants, and small planets within 2–3 sigma.
The PASTA II survey of 25 stars with high-resolution ground-based spectra and 23 elements confirmed it: the Sun is depleted relative to all of them without exception, regardless of planet type, and only among hosts of small planets is the depletion slightly stronger—at the level of a hint, not significance. A separate 2025 paper on 50 stars with 29 giant planets added that the slopes correlate more strongly with a star’s metallicity, and stated the conclusion outright: the peculiarity of the Sun’s composition is primarily tied to Galactic chemical evolution.
And here is the rough edge I would not smooth out. That same year, a paper appears on 99 twins and analogs, where a neural network reconstructs 20 elements, and it claims the opposite: the Sun is more depleted than 89 % of the analogs studied, while the difference from the subsample of stars without detected exoplanets reaches 9.5 sigma. In other words, one half of the field says, “the effect exists, and it is connected to planets,” while the other says, “the population-level test cannot see it.” The authors of the third paper take the sample from that very article with its 9.5 sigma and show that 64 of the 99 stars fit the prediction of their own Galactic model within one sigma. Both sides can be right only in one sense: a chemical signal of planetary origin may exist, but be drowned out by the scatter created by the Galaxy itself.
The actual turn in 2025 is that researchers stopped treating the condensation-temperature slope as a property of the Sun and began treating it as a property of the elements. Under the new framework, each star is described not by age and metallicity, but by two numbers: what share of its heavy elements came from core-collapse supernovae, and what share came from thermonuclear Type Ia supernovae. It turned out that elements with high condensation temperatures systematically receive more from the former. Which means the correlation of abundances with condensation temperature is not an imprint of a dust trap or a swallowed planet, but an imprint of nucleosynthesis: different stars are assembled from different supernova mixes, and the Galaxy is heterogeneous here on scales we once considered too small.
The scale of the test is entirely different from the twin-star papers: 109,500 solar analogs from four surveys, including Gaia, GALAH, and APOGEE spectra. Metallicity and the ratio of alpha elements to iron predict the slope with explained variances of 15 and 23 %, and within this framework the Sun looks ordinary: its profile corresponds to a star that is richer than average by 0.1 dex in iron and depleted in alpha elements by 0.02 dex. Citing a 2021 mixture-distribution model, the authors note that the “80 %” from older papers is the share of stars with a positive slope relative to zero, not relative to the population: if the sample is split into two components, the Sun lands in the center of the numerous group, not at its edge.
The test for planetary ingestion in the same paper is another good detail. The “eaten planet” effect from Nature 2024 (91 pairs of coeval stars that grew up together, at least seven cases of ingestion, around 8 % across the population) was searched for pairwise among 109,500 stars. Profiles matching that signature occurred exactly as often as randomly generated ones. In other words, planetary ingestion is real, but at current precision it is indistinguishable from ordinary Galactic scatter.
Running in parallel with this argument is another one, far more unpleasant, and it is about how we still do not know the exact abundance of elements in the most familiar star we have. The solar abundance problem looks like this: the old 1996 tables gave the fraction of all elements heavier than helium as Z = 0.017, and such a solar model matched helioseismology beautifully—the depth of the convective zone and the sound-speed profile, measured from surface oscillations. A refined analysis in 2005 dropped the oxygen abundance from 8.83 to 8.66 and Z to 0.0122—and the model stopped agreeing with the observations. Since then, astrophysicists have lived in reconciliation mode: more careful three-dimensional models and the inclusion of non-equilibrium effects raised Z back to 0.0139, other groups hold to oxygen values of 8.73–8.76, while attempts to save the matter through late accretion of metal-poor gas improve some profiles and spoil others.
Gustafsson’s 2025 review ends more honestly than popular texts usually do: there is no solution. Planetary versions, Galactic migration, a local supernova explosion, radiative “cleansing” of dust in a star-forming region, turbulent gas-and-dust separation, self-cleansing during slow accretion, atomic diffusion inside the star itself—all explanations either produce half the amplitude, contradict the chronology, or fail to hold up in samples. The author opens by quoting Holmes (“when you have eliminated the impossible...”) and at the end admits that the step-by-step procedure of elimination could not be repeated: high-quality data expose a structure that demands models more complicated than the mystery itself.
One detail remains, and I like it more than the effect itself. The Sun is formally poor in the refractory belt, but in biogenic elements—carbon, nitrogen, oxygen, sulfur, phosphorus—it is, on the contrary, richer than its twins by roughly the same 20 %, because those are volatiles. The original 2007 observing-time proposal was titled “The Basic Building Blocks of Life in Solar Twins,” and the result came out exactly as the title promised, although specialists see no direct connection between a 20 % phosphorus surplus and the existence of life: no hard thresholds for biogenic elements are known, and phosphorus and sulfur in biochemistry appear to be interchangeable.
As for the anomaly of the system as a whole, an appealingly mundane picture emerges. The Sun is slightly richer in metals than its contemporaries, noticeably quieter in activity than other stars of the same type, poorer in lithium than it ought to be for its age, and, according to the best models, not particularly exceptional at all when viewed through the right statistics. An ordinary star in a queue of one hundred nine thousand—and that is exactly why it is so difficult.