The Ashes of Supermassive Stars: Globular Cluster-like Aluminum Enhancement in Little Red Dots

V. Kokorev, J. Chisholm, R. P. Naidu, M. Gieles, S. Finkelstein, D. Berg, H. Akins, A. Taylor, S. Fujimoto, L. J. Furtak, J. Greene, A. de Graaff, K. Hawkins, T. Hsiao, D. Nandal, J. Matthee, S. Monty, P. Rinaldi, M. Boylan-Kolchin

First listed 2026-09-10 | Last updated 2026-09-08

Abstract

The relative abundances of elements in galaxies serve as fossil records of the physical conditions and processes by which they were forged. While the Big Bang produced only the lightest elements, subsequent stellar nucleosynthesis imprinted characteristic abundance patterns onto the surrounding gas, set initially by the temperatures reached inside stars and subsequently shaped by how the processed material was mixed and released. Globular clusters - dense, ancient groups of stars - provide a striking unique example. Some contain stars depleted in magnesium and enriched in aluminum, showing that they formed from gas exposed to exceptionally hot hydrogen burning. The stars responsible remain unknown. Little Red Dots may provide this missing engine. These compact, luminous objects formed at cosmic epochs similar to those associated with globular-cluster formation and are enshrouded by dense gas whose chemical composition can be measured with the James Webb Space Telescope. Here, using deep spectroscopy from the SPURS program, we show that this abundance pattern characterizes the LRD central engine: magnesium-depleted and aluminum-enhanced gas with a metallicity only 1% that of the Sun. This pattern is not produced by ordinary massive stars at these redshifts and cannot be mimicked by ionization, gas geometry or dust. Instead, it is reproduced by hot hydrogen burning in fully convective supermassive stars, with the measured abundances implying masses of at least 10,000 solar masses - approximately 100 times larger than any star observed in the present-day Universe. Little Red Dots may therefore reveal supermassive stars during their brief lives or in the immediate aftermath of their direct collapse, simultaneously identifying the long-sought source of globular cluster abundance anomalies and a formation pathway for massive black hole seeds.

Short digest

Using deep SPURS JWST/NIRSpec rest-UV spectroscopy of four bright little red dots, Kokorev et al. derive Mg, Al, Si, and Fe abundances from resolved absorption features in dense gas surrounding the central engines. The LRDs are metal poor, at roughly 1% solar metallicity, yet show an extreme globular-cluster-like signature of magnesium depletion and aluminum enhancement while retaining near-solar Si/Fe, unlike matched star-forming galaxies and conventional massive-star enrichment. The inferred 73–81 MK hot-hydrogen-burning conditions are consistent with fully convective supermassive stars of at least 10,000 solar masses, suggesting LRDs may expose these stars shortly before or after direct collapse. This provides a potential common origin for globular-cluster light-element anomalies and massive black-hole seeds.

Key figures to inspect

• Figure 1. Shows the four compact SPURS targets and the inverse-variance-weighted rest-UV stack from which Mg, Al, Si, and Fe absorption measurements are extracted. It establishes both the sample definition and the observational basis for the abundance analysis.
• Figure 2. This is the paper's central empirical result: the LRDs occupy the extreme Al-rich, Mg-poor extension of the globular-cluster abundance sequence while remaining strongly offset from matched high-redshift and local star-forming galaxies. The Si-versus-Al comparison further shows that the anomaly is not a generic metal-enrichment effect.
• Figure 3. Contrasts alpha-capture enrichment with the hot-hydrogen-burning pathway that converts Mg into Al in fully convective stars. It provides the physical interpretation for why the observed combination of Mg depletion, Al enhancement, and limited Si production is diagnostic.
• Figure 4. Connects the measured Mg, Al, and Si ratios to nucleosynthesis tracks and then maps the inferred core-temperature range onto stellar mass. This synthesis figure carries the conclusion that the polluting source must be a supermassive star with a mass of at least 10,000 solar masses.