The complex stellar system M 22: constraining the chemical enrichment from AGB stars using magnesium isotope ratios

Astronomers studying the chemical makeup of stars in the M 22 cluster have uncovered a hidden timeline written in magnesium isotopes. By analyzing high-resolution light spectra, the researchers found that low-mass dying stars—specifically asymptotic giant branch (AGB) stars—likely enriched the cluster with heavy elements. Crucially, the specific isotopic signature of magnesium suggests that the two distinct groups of stars in M 22 are separated by an age difference of at least 280 to 480 million years.

The unresolved history of M 22

Galactic globular clusters are typically viewed as simple stellar populations. They are usually groups of stars born at the same time from the same gas cloud. However, many clusters exhibit "multiple populations" with unexpected chemical variations. M 22 (NGC 6656) is an extreme example of this complexity. It is a "Type II" cluster. This means it displays a spread in iron-peak elements and neutron-capture elements.

Specifically, M 22 contains two distinct groups of stars: one "s-process rich" and one "s-process poor." The s-process (slow neutron-capture process) is a nucleosynthetic pathway. In this process, atomic nuclei capture neutrons one by one to build heavier elements like yttrium and lanthanum. For years, astronomers hypothesized that the s-process rich group was "polluted" by the ejecta of AGB stars. These are intermediate-mass stars in their final evolutionary stages. Previous models suggested these polluters were massive, between 3 and 6 solar masses ($M_\odot$). However, these models faced a critical tension. If these massive AGB stars were the source, they should have significantly altered the ratios of magnesium isotopes. Specifically, they would have changed the levels of $^{25}\text{Mg}$ and $^{26}\text{Mg}$. Current data cannot easily reconcile those massive models with the observed magnesium levels.

Decoding isotopes through MCMC optimization

To resolve this tension, the authors report the first-ever measurement of magnesium isotopic abundance ratios in a metal-poor, globular-cluster-like system. This required capturing extremely high-fidelity data. They used spectra with a resolution of $R = 110,000$ and a signal-to-noise ratio ($S/N$) of 300 per pixel. These were obtained using the UVES spectrograph on the Very Large Telescope (VLT).

The measurement process relies on identifying subtle asymmetries in the MgH (magnesium hydride) molecular lines. Because the neutron-rich isotopes $^{25}\text{Mg}$ and $^{26}\text{Mg}$ have different masses, their spectral lines are slightly shifted. This creates a characteristic "red wing" or asymmetry in the combined feature .

To extract these ratios accurately, the researchers moved away from traditional "eye-fitting" or simple grid searches. Such methods often struggle with parameter degeneracy (where different physical settings produce nearly identical spectral shapes). Instead, they developed ratio (Rapid, AuTomatic Isotope Optimisation), a custom Python-based wrapper for the MOOG spectral synthesis code.

The ratio tool employs a Markov Chain Monte Carlo (MCMC) ensemble sampler. This is a statistical method that explores a range of possibilities to find the most likely answer. It optimizes six distinct parameters simultaneously: 1. Total magnesium abundance ($\log\epsilon(\text{Mg})$). 2. Macroturbulent broadening ($S$), which accounts for large-scale gas motions in the stellar atmosphere. 3. The $^{25}\text{Mg}/^{24}\text{Mg}$ ratio. 4. The $^{26}\text{Mg}/^{24}\text{Mg}$ ratio. 5. Continuum correction (adjusting the baseline of the spectrum). 6. Radial velocity correction (accounting for the star's motion toward or away from Earth).

By using an MCMC approach, the authors can map the "posterior distribution." This is the probability landscape of these parameters. It allows them to identify the most likely isotopic ratios. It also provides realistic, mathematically rigorous error bars .

Constraints from s-process and light elements

The results provide a sharp pivot in our understanding of M 22's enrichment. The authors find that while there are star-to-star variations in magnesium isotopes, there is no significant correlation between these isotopes and the heavy s-process elements. This implies that the mechanism producing the heavy elements is decoupled from the mechanism altering the magnesium isotopes.

Instead, the study identifies a strong correlation between the $^{26}\text{Mg}/^{24}\text{Mg}$ ratio and light elements like sodium (Na) and aluminum (Al). It also shows an anti-correlation with oxygen (O) [, Figure 6].

Specifically, stars with higher $^{26}\text{Mg}$ tend to be Na-enhanced and O-poor.

The most consequential finding comes from comparing these isotopic ratios to custom AGB nucleosynthesis models. The authors report that the s-process rich population shows only a minimal difference in magnesium isotopes compared to the s-process poor population. They found an average difference of $\sim 0.033$ for $^{25}\text{Mg}/^{24}\text{Mg}$ and $\sim 0.024$ for $^{26}\text{Mg}/^{24}\text{Mg}$. When these observed values are mapped against AGB models, the data strongly favor low-mass AGB stars in the range of $1\text{--}3\ M_\odot$. High-mass models (above $3.5\ M_\odot$) cannot reproduce the observed s-process enhancement in elements like lanthanum (La) and neodymium (Nd) without over-predicting the magnesium isotope shifts [, Figure 11].

This mass constraint leads to a direct temporal conclusion. Since $1\text{--}3\ M_\odot$ stars live significantly longer than $3\text{--}6\ M_\odot$ stars, the "polluter" stars must have taken much longer to evolve. This implies an age gap of at least 280 to 480 million years between the two stellar populations.

Limitations and systematic uncertainties

Despite the high precision of the ratio method, several caveats remain. First, the sample size is small. It consists of only six stars (three from each group). While this allowed for high-quality individual spectra, it limits the statistical power to detect subtler correlations.

Second, the study acknowledges significant systematic uncertainties in AGB modeling. Factors such as the treatment of convection (the movement of gas within the star), mass-loss rates, and specific nuclear reaction rates can alter the predicted yields of isotopes. Consequently, the inferred age difference of 280–480 Myr depends on the specific AGB models used for the analysis.

Finally, the authors note that potential Non-Local Thermodynamic Equilibrium (NLTE) effects might exist. These occur when matter and radiation in the stellar atmosphere are not in perfect balance. Such effects might lead to an underestimation of the actual line depths. This could subtly bias the isotopic derivations.

The verdict on M 22's formation

The evidence points toward a low-mass enrichment scenario. By using the isotopic signatures of magnesium as a "clock" and a "thermometer" for stellar interiors, the authors have effectively ruled out the high-mass AGB models previously favored for M 22.

The verdict is a qualified yes regarding the AGB enrichment hypothesis. However, there is a major revision of the progenitor characteristics. The enrichment was likely driven by low-mass stars ($1\text{--}3\ M_\odot$). This necessitates a prolonged period of star formation. Or, it suggests a secondary burst of star formation occurred hundreds of millions of years after the initial cluster assembly. Future research must now focus on whether this low-mass AGB signature is a universal feature of complex globular clusters or a specific quirk of M 22's unique environment.

Figures from the paper