The chemical make-up of the SunA 2020 vision

For most of the twentieth century, the Sun's chemical composition was treated as settled. Astronomers had a table of numbers — the abundances of every element in the solar atmosphere — and they used it as the universal yardstick for everything from newborn stars to distant galaxies. Then, in the early 2000s, a new generation of solar models tore that table apart. Not with a dramatic discovery, but with better physics. The numbers that came out didn't match the Sun's own interior. We remeasured the nearest star, and suddenly we couldn't explain it. This paper — Asplund, Amarsi, and Grevesse's 2020 reassessment — is the field's most comprehensive attempt yet to resolve that tension. The reason solar abundances matter so broadly is simple: they are the reference point. When astronomers characterize the chemistry of a star, a nebula, or a galaxy, they express it relative to the Sun. Stellar evolution models, galactic chemical evolution simulations, nucleosynthesis calculations — all anchored to these numbers. The scale itself is logarithmic, with hydrogen fixed by definition at twelve. So when Asplund and colleagues report carbon at a logarithmic epsilon of eight point four six, they mean: take the ratio of carbon atoms to hydrogen atoms, take the base-ten logarithm, and add twelve. Each unit on this scale is a factor of ten in actual abundance. The compilation that became the community standard was Anders and Grevesse in 1989, now among the ten most cited papers in astronomy, with over eight thousand citations. The Asplund team's 2020 reassessment covers all eighty-three long-lived elements and is the most thorough update since. What makes the new analysis different starts with the atmosphere model. For most of the twentieth century, solar spectroscopy used one-dimensional, static model atmospheres and assumed local thermodynamic equilibrium — LTE — meaning that atomic level populations follow simple statistical distributions set by local temperature and pressure. That assumption is convenient. It's also wrong in ways that matter. In the real solar photosphere, photons are escaping into space, radiation fields are anisotropic, and atomic energy level populations are driven away from their equilibrium values. When you assume LTE and it isn't, the abundances you infer are systematically off. Asplund and colleagues use the Stagger code to build a fully three-dimensional radiative-hydrodynamical model of the solar surface. In this model, convection isn't parameterized — it emerges. The model's effective temperature over a twenty-four hour simulation sequence is five thousand seven hundred seventy-three kelvin, essentially identical to the measured solar value of five thousand seven hundred seventy-two kelvin. Critically, because the turbulent velocities are physically realistic, the analysis requires no ad hoc fudge factors to match observed line widths. The old one-dimensional approach needed free parameters called micro- and macroturbulence to reproduce line profiles. The three-dimensional model doesn't. That means the atmospheric structure is genuinely self-consistent. On top of the three-dimensional atmosphere, the team runs non-LTE line formation using a code called Balder. This solves for the actual level populations by requiring steady-state statistical equilibrium — the rate of transitions into each atomic level equals the rate out — rather than assuming the equilibrium is thermal. For thirteen elements, including carbon, nitrogen, oxygen, sodium, magnesium, and iron, the analysis is fully three-dimensional non-LTE. The combined effect of realistic atmospheric structure and proper non-LTE physics pulls the inferred abundances down, particularly for the three elements that matter most. Carbon, nitrogen, and oxygen are the headline results. They dominate the Sun's metal content, control stellar opacities, and set the pace of nuclear burning in stars. The new values are: a logarithmic epsilon of carbon equals eight point four six, plus or minus zero point zero four; nitrogen, seven point eight three, plus or minus zero point zero seven; oxygen, eight point six nine, plus or minus zero point zero four. What's striking is how they were obtained. The analysis draws on eight hundred seventy-nine molecular transitions — lines from CH, C2, CO, NH, CN, and OH — alongside atomic lines treated with full non-LTE physics. Those molecular transitions probe completely different formation conditions than the atomic lines. They respond differently to temperature, pressure, and radiation. Yet they converge on the same abundances. That agreement across such different physics is a powerful validation. Not every element can be measured in the photospheric spectrum. Noble gases — neon, argon, krypton, xenon — leave no detectable photospheric lines. For those, Asplund and colleagues turn to solar wind data returned by NASA's Genesis mission, which captured actual solar material and brought it back to Earth. Neon comes in at a logarithmic epsilon of eight point zero six; argon, six point three eight; krypton, three point one two; xenon, two point two two. Helium is derived not from spectroscopy or solar wind, but from helioseismology — the study of pressure waves resonating through the Sun's interior — giving a logarithmic epsilon of ten point nine one. Fifteen additional elements are taken from primitive CI chondrite meteorites, the most chemically unaltered material in the solar system. The meteorite comparison is a powerful consistency check. For the fifty-eight non-volatile elements where both photospheric spectroscopy and meteoritic measurements are available, the weighted mean difference is essentially zero — zero point zero zero zero plus or minus zero point zero zero seven dex, with a scatter of just zero point zero five dex. Two completely independent methods — one pointing a telescope at the Sun and one dissolving rocks in a laboratory — agree on the composition of the solar system to better than ten percent. There's also a subtler pattern hiding in that comparison. Moderately volatile elements show a small systematic enrichment in meteorites relative to the Sun, and when you plot the difference against condensation temperature, a gentle linear trend emerges — possibly a signature of planet formation in the early solar system. Now for the problem. The lower carbon, nitrogen, and oxygen abundances reduce the Sun's overall metallicity — the mass fraction of elements heavier than helium — to Z equals zero point zero one three nine, with Z over X coming out at zero point zero one eight seven. Older compilations gave Z over X values around zero point two three to zero point two seven. That's not a small shift. And when you feed the new lower metallicity into a standard solar model, the model's predictions for the Sun's interior stop matching what helioseismology measures. Helioseismology is essentially the Sun's own MRI. Pressure waves — called p-modes — resonate through the solar interior, and their frequencies encode the sound speed at every depth, the location of the boundary between the convective and radiative zones, and the helium abundance in the convection zone. With the old high-metallicity abundances, standard solar models matched all three of these seismic diagnostics well. With the new low-metallicity values, the agreement breaks down significantly. The predicted convection zone depth is wrong. The surface helium abundance is wrong. The sound speed profile is wrong. The Sun, measured at its surface, disagrees with the Sun probed by its sound waves. Two fixes have been proposed. One is that the atomic opacity calculations used in solar interior models are underestimated — resolving the mismatch would require opacities to be about twenty percent higher near the base of the convection zone. Recent laboratory experiments do find higher opacities than models predict, with iron alone accounting for roughly half the needed increase. The other fix involves additional mixing below the convection zone. Neither resolves the discrepancy fully. The tension is real and unresolved. What Asplund, Amarsi, and Grevesse have produced is not a final answer. They are explicit about that. Only twenty-six elements have been treated in non-LTE, and many only in one dimension. The proto-solar abundances — corrected upward by about zero point zero six dex in metals for the effects of atomic diffusion over four point six billion years — give an initial metal mass fraction of about zero point zero one five four. That number, along with the full table of eighty-three elements, is what the next generation of stellar and galactic models will use. The nearest star, scrutinized spectroscopically for more than a century, still hides subtle signatures in its light. Those signatures, it turns out, are also the keys to understanding its interior — and that is a genuinely strange place for solar physics to find itself, with better measurements and a deeper puzzle. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.