Supernova in a Globular Cluster: How M92 Preserved the Traces of an Explosion ✦
Globular clusters exhibit unique light-element abundance patterns that cannot be explained by standard enrichment mechanisms: first-population stars are sodium-poor, while second-population stars are sodium-rich and found only in clusters. A differential line-by-line analysis of stars in M92 within a narrow range of effective temperatures revealed that the second population (with elevated sodium) has a higher iron level than the first. The populations are clearly separated in Na, Al, and Fe content. This iron enhancement indicates that M92 retained at least some of the supernova ejecta, all of which exploded after the first population had finished forming. The result sets a lower bound on the time delay between star formation episodes and challenges the long-held view that globular clusters are incapable of chemical evolution driven by supernovae.
Context
The origin of multiple stellar populations in globular clusters is one of the greatest mysteries in modern astrophysics. Unlike galaxies, these dense stellar swarms don't exhibit the usual chemical evolution driven by supernova explosions. Instead, they're famous for anticorrelations of light elements like sodium and oxygen, thought to arise from hydrogen burning at high temperatures inside massive stars. These nucleosynthesis reactions were described in detail by Hans Bethe. However, the connection between these processes and the formation of iron-peak elements—and thus the timeline of star formation—remained unclear.
Methods
The researchers applied a differential spectral analysis method, comparing 11 red giant branch stars in M92 selected from a very narrow range of colors and luminosities. This minimized systematic errors from atmospheric models and oscillator strengths. Spectra were obtained with the Keck telescope using the HIRES spectrograph at a resolution of about 45,000. For each line, the abundance difference relative to the cluster mean was computed, then averaged over all lines of the element. This approach captured tiny differences in iron content—around 0.08 dex—between populations, undetectable with ordinary methods. A key role was played by using photometry from Gaia to determine temperatures and applying the Stefan-Boltzmann law to calculate surface gravity.
Results
In M92, stars with high sodium content (the second population) were found to have higher abundances of iron and other iron-peak elements compared to the sodium-poor first population. The difference is 0.082 ± 0.016 dex, with a permutation test yielding a significance level of p = 0.002. Moreover, the first population shows an extremely low iron scatter (σ < 0.038 dex), while the second spans the full range of elevated values. Meanwhile, magnesium—burned in hydrogen cycles—anticorrelates with sodium and doesn't follow a simple dilution model, indicating a prolonged enrichment process lasting more than 3 million years—the typical time before the most massive supernovae explode, as predicted by Fritz Zwicky. Additionally, a previously noted dispersion in r-process elements (especially europium) in the first population was confirmed—but not in the second—hinting at a possible contribution from neutron stars to the early enrichment of the protocluster.
Implications
This result directly proves that the globular cluster M92 retained at least some of the ejecta from core-collapse supernovae that exploded after the first generation had already formed. Thus, the time interval between generations is at least 3 million years. This delay rules out fast-rotating massive stars as the main source of light elements in this cluster and challenges models with extremely massive stars. The discovery also shows that even 'ordinary' globular clusters can preserve supernova signatures if analyzed with sufficient precision, forcing a reexamination of early stellar evolution in dense environments.
Future development
If this differential analysis is applied to other globular clusters, we could determine whether M92 is unique or if many clusters retained supernovae in their early stages. Future observations with even higher spectral resolution and in the infrared (where non-LTE effects are smaller) will allow measurements of other elements like potassium and aluminum with better accuracy. Combining these data with direct helium and lithium abundance measurements is also promising for reconstructing the full picture of nucleosynthesis in protoclusters.
Impact
The findings will influence the theory of globular cluster formation and evolution, models of chemical evolution in the early universe, and the interpretation of data from the James Webb Space Telescope, which observes similar processes in very distant stellar systems.
Next steps
Immediate next steps include expanding the sample of stars in M92 to other evolutionary stages (e.g., horizontal branch stars) and conducting similar analyses for other clusters, including anomalous ones, to uncover general patterns in supernova retention.
Key open problems
This discovery links two key unsolved problems: the origin of multiple populations in globular clusters and the puzzle of why most lack supernova enrichment. It also touches on the issue of r-process sources, confirming that an r-process event occurred in the proto-cloud before the first generation formed, without leaving any iron signature.
🎯 Globular cluster M92, one of the oldest in the Galaxy, lies in the constellation Hercules and hosts around 330,000 stars. Despite being over 12 billion years old, its stars still carry a 'chemical memory' of the first few million years—like the growth rings of a tree.
Key numbers
- iron abundance difference: 0.082 dex
- significance level: 0.002
- lower limit on delay time: 3 million years
- sodium gap between populations: 0.450 dex
- upper limit on iron scatter in the first population: 0.038 dex