A simulation burning 40 million processor hours modelled how the first stars seeded early galaxies with iron

The MEGATRON project spent 40 million processor hours on the United Kingdom’s national supercomputers, a budget the University of Bath likens to five million laptops running at once for a year. The payoff so far is four studies in the Open Journal of Astrophysics that follow how the universe’s first stars, and the supernovae that ended their lives, spread new chemical elements into the gas that later formed galaxies.

The result is a simulation, not a measurement. It shows what physics produces when it is run forward from pristine gas, and it earns trust only where its output matches what telescopes and stellar surveys see today. Despite the popular framing of tracing the origin of the atoms in a human body, the paper behind the project reports on iron, and no breakdown of carbon by source has been published in the material reviewed.

Pair-instability supernovae and the iron plateau

The study the University of Bath release cites is “MEGATRON: how the first stars can create an iron metallicity plateau in the smallest dwarf galaxies,” published on September 30, 2026 (DOI 10.33232/001c.169605) in The Open Journal of Astrophysics. Martin Rey and Harley Katz are the first two of 26 listed authors, a group that also includes researchers at the University of Chicago and the Institut d’Astrophysique de Paris, the project’s partner institutions.

The central finding concerns the smallest dwarf galaxies. In the simulated population, an overabundance of dwarfs sit on a flat plateau at an iron abundance of about [Fe/H] of minus 2.5, a logarithmic scale on which each step down means ten times less iron relative to hydrogen than the Sun has, at stellar masses of 100,000 solar masses or less. The authors link that plateau to enrichment by Population III pair-instability supernovae, the extremely violent explosions of very massive first-generation stars described in general terms in a reference entry on the process.

A strong Lyman-Werner background of ultraviolet radiation from the protogalaxy, the paper explains, means such explosions occur in dark matter haloes of about 10 million solar masses. Those haloes are massive enough to hold onto the ejected material instead of losing it to space, so the iron stays and enriches the next stars. About 20 percent of simulated dwarfs are predicted to be iron-deficient, at [Fe/H] of minus 3 or lower.

MEGATRON runs: pristine gas, radiation and a 40-million-hour budget

The University of Bath describes the 40 million processor hours on the United Kingdom’s national supercomputers as the computing budget for the whole project. According to the Bath release as carried by ScienceDaily, the runs begin with pristine gas containing no heavy elements, meant to resemble conditions shortly after the Big Bang. They then follow the formation of the first stars, their radiation, the supernovae at the end of their lives, and the dispersal of new elements into later stars and galaxies, tracking gas motion, the travel of starlight and the changing abundance of each element. The target region evolves toward a galaxy of roughly Milky Way mass.

The paper describes these as cosmological radiation-hydrodynamics simulations run to a redshift of about 8, with Population III and Population II star formation, feedback and chemical enrichment modelled self-consistently. The release does not name the individual machines, and it gives no resolution or parameter values, so the 40 million hours cannot be tied to a named facility from the sources read here.

Radiation is the expensive part. The team argues that simpler models may underestimate how strongly stellar radiation and chemical processes shape the gas around young galaxies, and that high resolution captures gas structures such approaches miss.

Simulated dwarfs checked against redshift-zero metallicity

Validation comes from the present day. The paper states that the simulated dwarfs agree with the observed relation between stellar mass and metallicity at redshift zero, and it adds that both the plateau and the iron-poor tail survive large variations in the assumptions about Population II feedback and persist in bound satellites of the central galaxy. That is a robustness test of the model against its own knobs, which is useful but different from a direct observation of a Population III explosion, none of which has been seen.

Rey, a physicist at the University of Bath and a lead contributor, framed the stakes in terms of two kinds of evidence that rarely meet. The release says accurate modelling is needed to connect James Webb Space Telescope observations of young galaxies with chemical fingerprints preserved in ancient stars, and in Rey’s words, “MEGATRON provides a physical bridge between the two.”

What remains contested about carbon

The claim that the project traced the origin of carbon is the part that does not hold up against the documents. The paper addresses iron, and the release names no carbon or oxygen fractions, no initial mass function and no error bars. Any statement of what share of the carbon in a person came from first-generation stars, ordinary supernovae or older, evolved stars would be an inference layered on a model, and the material reviewed does not supply one.

The wider caveat is that the first-star story depends on assumptions that cannot yet be pinned down. The existence of the plateau, the halo mass at which pair-instability explosions retain their ejecta and the fraction of iron-poor dwarfs are predictions that observers can test against surveys of faint dwarf galaxies. Those tests are the next step, and the release frames the project as running through 2030, with more results to come.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

More from Morning Overview